We present Y-zipper, a novel three-sided 3D-printed zipper structure that enables three flexible strips to interlock and transform into a rigid rod-like form. Building on this flex–rigid transition mechanism, we further design a specialized slider to achieve rapid and reversible zipping interactions. This slider serves as the basis for three actuation methods—manual, dynamic mechanical, and static mechanical—which enable both remote control and automated closure and release. In addition, Y-zipper provides four motion primitives: straight, bend, coil, and screw, whose combinations extend the flex–rigid transition mechanism to spatial curve structures. To support customization, we develop a computational design tool that automatically generates zipper geometry based on input primitives, unfolds the structure for 3D printing, and embeds both teeth and compliant bridges. Controlled experiments evaluate its mechanical properties, repeatability, and actuation speed, demonstrating robustness and reliability. Finally, we showcase a series of functional prototypes, including a medical wrist brace, a kinetic art installation, and a rapidly deployable tent structure.

We present VisiPrint, a tool for appearance-first previews of 3D-printed objects. Existing print preview slicers focus on toolpaths, not appearance, while pure rendering software is complex and cannot automatically reproduce slicing patterns. Prior work highlights persistent gaps between digital previews and printed results, such as color shifts, gloss/translucency changes, and layer-line highlights, motivating the creation of VisiPrint, an appearance-focused support tool. The VisiPrint algorithm combines slicer screenshots with filament photos via a custom diffusion-based synthesis pipeline. We present both a standalone user interface for VisiPrint compatible with any slicer and an Ultimaker Cura Plugin. We evaluate VisiPrint through a user study showing it is significantly faster, easier to use, and more faithful than alternatives: within a time-limit, participants completed 100% of preview tasks with VisiPrint, versus 63% with Cura and 13% with Blender. VisiPrint narrows the gap between design intent and printed appearance, complementing settings-centric tools with appearance-driven decision support.

We propose a method for generating fly-through videos of a scene, from a single image and a given camera trajectory. We build upon an image-to-video latent diffusion model [5]. We condition its UNet [25] denoiser on the camera trajectory, using four techniques. (1) We condition UNet's temporal blocks on raw camera extrinsics, similar to MotionCtrl [36]. (2) We use images containing camera ray parameters, similar to CameraCtrl [14]. (3) We re-project the initial image to subsequent frames and condition on the resulting video. (4) We introduce a global 3D representation using 2D ⇔ 3D transformers [32], which implicitly conditions on the camera poses. We combine all conditions in a ContolNet-style [42] architecture. We then propose a metric that evaluates overall video quality and the ability to preserve details with view changes, which we use to analyze the trade-offs of individual and combined conditions. Finally, we identify an optimal combination of conditions. We calibrate camera positions in our datasets for scale consistency across scenes, and we train our scene exploration model, CamCtrl3D, demonstrating state-of-the-art results.

Despite the unprecedented progress in the field of 3D generation, current systems still often fail to produce high-quality 3D assets that are visually appealing and geometrically and semantically consistent across multiple viewpoints. To effectively assess the quality of the generated 3D data, there is a need for a reliable 3D evaluation tool. Unfortunately, existing 3D evaluation metrics often overlook the geometric quality of generated assets or merely rely on black-box multimodal large language models for coarse assessment. In this paper, we introduce Eval3D, a fine-grained, interpretable evaluation tool that can faithfully evaluate the quality of generated 3D assets based on various distinct yet complementary criteria. Our key observation is that many desired properties of 3D generation, such as semantic and geometric consistency, can be effectively captured by measuring the consistency among various foundation models and tools. We thus leverage a diverse set of models and tools as probes to evaluate the inconsistency of generated 3D assets across different aspects. Compared to prior work, Eval3D provides pixel-wise measurement, enables accurate 3D spatial feedback, and aligns more closely with human judgments. We comprehensively evaluate existing 3D generation models using Eval3D and highlight the limitations and challenges of current models. Project page: http://eval3d.github.io.

We introduce RandAR, a decoder-only visual autoregressive (AR) model capable of generating images in arbitrary token orders. Unlike previous decoder-only AR models that rely on a predefined generation order, RandAR removes this inductive bias, unlocking new capabilities in decoder-only generation. Our essential design enables random order by inserting a "position instruction token" before each image token to be predicted, representing the spatial location of the next image token. Trained on randomly permuted token sequences - a more challenging task than fixed-order generation, RandAR achieves comparable performance to its conventional raster-order counterpart. More importantly, decoder-only transformers trained from random orders acquire new capabilities. For the efficiency bottleneck of AR models, RandAR adopts parallel decoding with KV-Cache at inference time, enjoying 2.5 × acceleration without sacrificing generation quality. Additionally, RandAR supports inpainting, outpainting and resolution extrapolation in a zero-shot manner. We hope RandAR inspires new directions for decoder-only visual generation models and broadens their applications across diverse scenarios. Our project page is at https://rand-ar.github.io/.

Current video diffusion models achieve impressive generation quality but struggle in interactive applications due to bidirectional attention dependencies. The generation of a single frame requires the model to process the entire sequence, including the future. We address this limitation by adapting a pretrained bidirectional diffusion transformer to an autoregressive transformer that generates frames on-the-fly. To further reduce latency, we extend distribution matching distillation (DMD) to videos, distilling 50-step diffusion model into a 4-step generator. To enable stable and high-quality distillation, we introduce a student initialization scheme based on teacher’s ODE trajectories, as well as an asymmetric distillation strategy that supervises a causal student model with a bidirectional teacher. This approach effectively mitigates error accumulation in autoregressive generation, allowing long-duration video synthesis despite training on short clips. Our model achieves a total score of 84.27 on the VBench-Long benchmark, surpassing all previous video generation models. It enables fast streaming generation of high-quality videos at 9.4 FPS on a single GPU thanks to KV caching. Our approach also enables streaming video-to-video translation, image-to-video, and dynamic prompting in a zero-shot manner. We release our code and pretrained models.

We present WonderWorld, a novel framework for interactive 3D scene generation that enables users to interactively specify scene contents and layout and see the created scenes in low latency. The major challenge lies in achieving fast generation of 3D scenes. Existing scene generation approaches fall short of speed as they often require (1) progressively generating many views and depth maps, and (2) time-consuming optimization of the scene geometry representations. Our approach does not need to generate multiple views, and it leverages a geometry-based initialization that significantly reduces optimization time. Another challenge is generating coherent geometry that allows all scenes to be connected. We introduce the guided depth diffusion that allows partial conditioning of depth estimation. WonderWorld creates connected and diverse 3D scenes, each generated in less than 10 seconds on a single A6000 GPU, enabling real-time user interaction and exploration. We release full code, software, and interactive demos in https://kovenyu.com/WonderWorld/.

As the field of representation learning grows, there has been a proliferation of different loss functions to solve different classes of problems. We introduce a single information-theoretic equation that generalizes a large collection of mod- ern loss functions in machine learning. In particular, we introduce a framework that shows that several broad classes of machine learning methods are precisely minimizing an integrated KL divergence between two conditional distributions: the supervisory and learned representations. This viewpoint exposes a hidden information geometry underlying clustering, spectral methods, dimensionality re- duction, contrastive learning, and supervised learning. This framework enables the development of new loss functions by combining successful techniques from across the literature. We not only present a wide array of proofs, connecting over 23 different approaches, but we also leverage these theoretical results to create state-of-the-art unsupervised image classifiers that achieve a +8% improvement over the prior state-of-the-art on unsupervised classification on ImageNet-1K. We also demonstrate that I-Con can be used to derive principled debiasing methods which improve contrastive representation learners.

Current vision systems typically assign fixed-length representations to images, regardless of the information content. This contrasts with human intelligence —and even large language models—which allocate varying representational capacities based on entropy, context and familiarity. Inspired by this, we propose an approach to learn variable-length token representations for 2D images. Our encoder-decoder architecture recursively processes 2D image tokens, distilling them into 1D latent tokens over multiple iterations of recurrent rollouts. Each iteration refines the 2D tokens, updates the existing 1D latent tokens, and adaptively increases representational capacity by adding new tokens. This enables compression of images into a variable number of tokens, ranging from 32 to 256. We validate our tokenizer using reconstruction loss and FID metrics, demonstrating that token count aligns with image entropy, familiarity and downstream task requirements. Recurrent token processing with increasing representational capacity in each iteration shows signs of token specialization, revealing potential for object / part discovery.

We propose RelitLRM, a Large Reconstruction Model (LRM) for generating high-quality Gaussian splatting representations of 3D objects under novel illuminations from sparse (4-8) posed images captured under unknown static lighting. Unlike prior inverse rendering methods requiring dense captures and slow optimization, often causing artifacts like incorrect highlights or shadow baking, RelitLRM adopts a feed-forward transformer-based model with a novel combination of a geometry reconstructor and a relightable appearance generator based on diffusion. The model is trained end-to-end on synthetic multi-view renderings of objects under varying known illuminations. This architecture design enables to effectively decompose geometry and appearance, resolve the ambiguity between material and lighting, and capture the multi-modal distribution of shadows and specularity in the relit appearance. We show our sparse-view feed-forward RelitLRM offers competitive relighting results to state-of-the-art dense-view optimization-based baselines while being significantly faster. Our project page is available at: https://relit-lrm.github.io/.

According to Algorithmic Information Theory (AIT), intelligent representations compress data into the shortest possible program while remaining predictive of its content—exhibiting low Kolmogorov Complexity (KC). In contrast, most visual representation learning systems assign fixed-length representations to all inputs, ignoring variations in complexity or familiarity. Recent adaptive tokenization methods address this by allocating variable-length representations but typically require test-time search over multiple hypotheses to identify the most predictive one. Inspired by KC principles, we propose a one-shot adaptive tokenizer, KARL, that predicts the appropriate number of tokens for an image in a single forward pass, halting once its approximate KC is reached. The token count serves as a proxy for the minimum description length. KARL performs comparably to recent adaptive tokenizers while operating in a one-pass manner. Additionally, we present a conceptual study showing a correlation between adaptive tokenization and core ideas from AIT. We demonstrate that adaptive tokenization not only aligns with KC but also reveals empirical signals approximating AIT concepts such as sophistication and logical depth. Finally, we analyze predicted image complexity and interestingness across axes such as structure vs. noise and in-distribution vs. out-of-distribution familiarity, highlighting alignment with human annotations.

Abstract Diffusion models excel at text-to-image generation, especially in subject-driven generation for personalized images. However, existing methods are inefficient due to the subject-specific fine-tuning, which is computationally intensive and hampers efficient deployment. Moreover, existing methods struggle with multi-subject generation as they often blend identity among subjects. We present FastComposer which enables efficient, personalized, multi-subject text-to-image generation without fine-tuning. FastComposer uses subject embeddings extracted by an image encoder to augment the generic text conditioning in diffusion models, enabling personalized image generation based on subject images and textual instructions with only forward passes. To address the identity blending problem in the multi-subject generation, FastComposer proposes cross-attention localization supervision during training, enforcing the attention of reference subjects localized to the correct regions in the target images. Naively conditioning on subject embeddings results in subject overfitting. FastComposer proposes delayed subject conditioning in the denoising step to maintain both identity and editability in subject-driven image generation. FastComposer generates images of multiple unseen individuals with different styles, actions, and contexts. It achieves 300 $$\times $$ × –2500 $$\times $$ × speedup compared to fine-tuning-based methods and requires zero extra storage for new subjects. FastComposer paves the way for efficient, personalized, and high-quality multi-subject image creation. Code, model, and dataset are available here ( https://github.com/mit-han-lab/fastcomposer ).

We present DenseAV, a novel dual encoder grounding architecture that learns high-resolution, semantically meaningful, and audio-visual aligned features solely through watching videos. We show that DenseAV can discover the “meaning” of words and the “location” of sounds without explicit localization supervision. Furthermore, it automatically discovers and distinguishes between these two types of associations without supervision. We show that DenseAV's localization abilities arise from a new multi-head feature aggregation operator that directly compares dense image and audio representations for contrastive learning. In contrast, many other systems that learn “global” audio and video representations cannot localize words and sound. Finally, we contribute two new datasets to improve the evaluation of AV representations through speech and sound prompted semantic segmentation. On these and other datasets we show DenseAV dramatically outperforms the prior art on speech and sound prompted semantic segmentation. DenseAV outperforms the current state-of-the-art, ImageBind, on cross-modal retrieval using fewer than half of the parameters. Project Page: https://aka.ms/denseav

We propose a method to control material attributes of objects like roughness, metallic, albedo, and transparency in real images. Our method capitalizes on the generative prior of text-to-image models known for photorealism, employing a scalar value and instructions to alter low-level material properties. Addressing the lack of datasets with controlled material attributes, we generated an object-centric synthetic dataset with physically-based materials. Finetuning a modified pretrained text-to-image model on this synthetic dataset enables us to edit material properties in real-world images while preserving all other attributes. We show the potential application of our model to material edited NeRFs.

Diffusion models generate high-quality images but require dozens of forward passes. We introduce Distribution Matching Distillation (DMD), a procedure to transform a diffusion model into a one-step image generator with minimal impact on image quality. We enforce the one-step image generator match the diffusion model at distribution level, by minimizing an approximate KL divergence whose gradient can be expressed as the difference between 2 score functions, one of the target distribution and the other of the synthetic distribution being produced by our one-step generator. The score functions are parameterized as two diffusion models trained separately on each distribution. Combined with a simple regression loss matching the large-scale structure of the multi-step diffusion outputs, our method outperforms all published few-step diffusion approaches, reaching 2.62 FID on ImageNet 64×64 and 11.49 FID on zero-shot COCO-30k, comparable to Stable Diffusion but orders of magnitude faster. Utilizing FP16 inference, our model can generate images at 20 FPS on modern hardware.

We introduce WonderJourney, a modular framework for perpetual 3D scene generation. Unlike prior work on view generation that focuses on a single type of scenes, we start at any user-provided location (by a text description or an image), and generate a journey through a long sequence of diverse yet coherently connected 3D scenes. We leverage an LLM to generate textual descriptions of the scenes in this journey, a text-driven point cloud generation pipeline to make a compelling and coherent sequence of 3D scenes, and a large VLM to verify the generated scenes. We show compelling, diverse visual results across various scene types and styles, forming imaginary “wonder journeys ”. Project website: https://kovenyu.com/WonderJourney/. “No, no! The adventures first, explanations take such a dreadful time.“ - Alice's Adventures in Wonderland

Deep features are a cornerstone of computer vision research, capturing image semantics and enabling the community to solve downstream tasks even in the zero- or few-shot regime. However, these features often lack the spatial resolution to directly perform dense prediction tasks like segmentation and depth prediction because models aggressively pool information over large areas. In this work, we introduce FeatUp, a task- and model-agnostic framework to restore lost spatial information in deep features. We introduce two variants of FeatUp: one that guides features with high-resolution signal in a single forward pass, and one that fits an implicit model to a single image to reconstruct features at any resolution. Both approaches use a multi-view consistency loss with deep analogies to NeRFs. Our features retain their original semantics and can be swapped into existing applications to yield resolution and performance gains even without re-training. We show that FeatUp significantly outperforms other feature upsampling and image super-resolution approaches in class activation map generation, transfer learning for segmentation and depth prediction, and end-to-end training for semantic segmentation.

Evaluating deep neural networks (DNNs) as models of human perception has given rich insights into both human visual processing and representational properties of DNNs. We extend this work by analyzing how well DNNs perform compared to humans when constrained by peripheral vision -- which limits human performance on a variety of tasks, but also benefits the visual system significantly. We evaluate this by (1) modifying the Texture Tiling Model (TTM), a well tested model of peripheral vision to be more flexibly used with DNNs, (2) generating a large dataset which we call COCO-Periph that contains images transformed to capture the information available in human peripheral vision, and (3) comparing DNNs to humans at peripheral object detection using a psychophysics experiment. Our results show that common DNNs underperform at object detection compared to humans when simulating peripheral vision with TTM. Training on COCO-Periph begins to reduce the gap between human and DNN performance and leads to small increases in corruption robustness, but DNNs still struggle to capture human-like sensitivity to peripheral clutter. Our work brings us closer to accurately modeling human vision, and paves the way for DNNs to mimic and sometimes benefit from properties of human visual processing.

Recent approaches have shown promises distilling expensive diffusion models into efficient one-step generators. Amongst them, Distribution Matching Distillation (DMD) produces one-step generators that match their teacher in distribution, i.e., the distillation process does not enforce a one-to-one correspondence with the sampling trajectories of their teachers. However, to ensure stable training in practice, DMD requires an additional regression loss computed using a large set of noise--image pairs, generated by the teacher with many steps of a deterministic sampler. This is not only computationally expensive for large-scale text-to-image synthesis, but it also limits the student's quality, tying it too closely to the teacher's original sampling paths. We introduce DMD2, a set of techniques that lift this limitation and improve DMD training. First, we eliminate the regression loss and the need for expensive dataset construction. We show that the resulting instability is due to the "fake" critic not estimating the distribution of generated samples with sufficient accuracy and propose a two time-scale update rule as a remedy. Second, we integrate a GAN loss into the distillation procedure, discriminating between generated samples and real images. This lets us train the student model on real data, thus mitigating the imperfect "real" score estimation from the teacher model, and thereby enhancing quality. Third, we introduce a new training procedure that enables multi-step sampling in the student, and addresses the training--inference input mismatch of previous work, by simulating inference-time generator samples during training. Taken together, our improvements set new benchmarks in one-step image generation, with FID scores of 1.28 on ImageNet-64×64 and 8.35 on zero-shot COCO 2014, surpassing the original teacher despite a 500X reduction in inference cost. Further, we show our approach can generate megapixel images by distilling SDXL, demonstrating exceptional visual quality among few-step methods, and surpassing the teacher. We release our code and pretrained models.

Creativity is an indispensable part of human cognition and also an inherent part of how we make sense of the world. Metaphorical abstraction is fundamental in communicating creative ideas through nuanced relationships between abstract concepts such as feelings. While computer vision benchmarks and approaches predominantly focus on understanding and generating literal interpretations of images, metaphorical comprehension of images remains relatively unexplored. Towards this goal, we introduce Meta-CLUE, a set of vision tasks on visual metaphor. We also collect high-quality and rich metaphor annotations (abstract objects, concepts, relationships along with their corresponding object boxes) as there do not exist any datasets that facilitate the evaluation of these tasks. We perform a comprehensive analysis of state-of-the-art models in vision and language based on our annotations, highlighting strengths and weaknesses of current approaches in visual metaphor classification, localization, understanding (retrieval, question answering, captioning) and generation (text-to-image synthesis) tasks. We hope this work provides a concrete step towards developing AI systems with human-like creative capabilities. Project page: https://metaclue.github.io

Motion magnification helps us visualize subtle, imperceptible motion. However, prior methods only work for 2D videos captured with a fixed camera. We present a 3D motion magnification method that can magnify subtle motions from scenes captured by a moving camera, while supporting novel view rendering. We represent the scene with time-varying radiance fields and leverage the Eulerian principle for motion magnification to extract and amplify the variation of the embedding of a fixed point over time. We study and validate our proposed principle for 3D motion magnification using both implicit and tri-plane-based radiance fields as our underlying 3D scene representation. We evaluate the effectiveness of our method on both synthetic and real-world scenes captured under various camera setups.

Priors are essential for reconstructing images from noisy and/or incomplete measurements. The choice of the prior determines both the quality and uncertainty of recovered images. We propose turning score-based diffusion models into principled image priors ("score-based priors") for analyzing a posterior of images given measurements. Previously, probabilistic priors were limited to handcrafted regularizers and simple distributions. In this work, we empirically validate the theoretically-proven probability function of a score-based diffusion model. We show how to sample from resulting posteriors by using this probability function for variational inference. Our results, including experiments on denoising, deblurring, and interferometric imaging, suggest that score-based priors enable principled inference with a sophisticated, data-driven image prior.

We present Muse, a text-to-image Transformermodel that achieves state-of-the-art image genera-tion performance while being significantly moreefficient than diffusion or autoregressive models.Muse is trained on a masked modeling task indiscrete token space: given the text embeddingextracted from a pre-trained large language model(LLM), Muse learns to predict randomly maskedimage tokens. Compared to pixel-space diffusionmodels, such as Imagen and DALL-E 2, Muse issignificantly more efficient due to the use of dis-crete tokens and requires fewer sampling itera-tions; compared to autoregressive models such asParti, Muse is more efficient due to the use of par-allel decoding. The use of a pre-trained LLM en-ables fine-grained language understanding, whichtranslates to high-fidelity image generation andthe understanding of visual concepts such as ob-jects, their spatial relationships, pose, cardinalityetc. Our 900M parameter model achieves a newSOTA on CC3M, with an FID score of 6.06. TheMuse 3B parameter model achieves an FID of7.88 on zero-shot COCO evaluation, along with aCLIP score of 0.32. Muse also directly enables anumber of image editing applications without theneed to fine-tune or invert the model: inpainting,outpainting, and mask-free editing. More resultsand videos demonstrating editing are available at https://muse-icml.github.io/

Denoising diffusion models are a powerful type of generative models used to capture complex distributions of real-world signals. However, their applicability is limited to scenarios where training samples are readily available, which is not always the case in real-world applications. For example, in inverse graphics, the goal is to generate samples from a distribution of 3D scenes that align with a given image, but ground-truth 3D scenes are unavailable and only 2D images are accessible. To address this limitation, we propose a novel class of denoising diffusion probabilistic models that learn to sample from distributions of signals that are never directly observed. Instead, these signals are measured indirectly through a known differentiable forward model, which produces partial observations of the unknown signal. Our approach involves integrating the forward model directly into the denoising process. A key contribution of our work is the integration of a differentiable forward model into the denoising process. This integration effectively connects the generative modeling of observations with the generative modeling of the underlying signals, allowing for end-to-end training of a conditional generative model over signals. During inference, our approach enables sampling from the distribution of underlying signals that are consistent with a given partial observation. We demonstrate the effectiveness of our method on three challenging computer vision tasks. For instance, in the context of inverse graphics, our model enables direct sampling from the distribution of 3D scenes that align with a single 2D input image.

We study the problem of extracting biometric information of individuals by looking at shadows of objects cast on diffuse surfaces. We show that the biometric information leakage from shadows can be sufficient for reliable identity inference under representative scenarios via a maximum likelihood analysis. We then develop a learning-based method that demonstrates this phenomenon in real settings, exploiting the subtle cues in the shadows that are the source of the leakage without requiring any labeled real data. In particular, our approach relies on building synthetic scenes composed of 3D face models obtained from a single photograph of each identity. We transfer what we learn from the synthetic data to the real data using domain adaptation in a completely unsupervised way. Our model is able to generalize well to the real domain and is robust to several variations in the scenes. We report high classification accuracies in an identity classification task that takes place in a scene with unknown geometry and occluding objects.

Separating an image into meaningful underlying components is a crucial first step for both editing and understanding images. We present a method capable of selecting the regions of a photograph exhibiting the same material as an artist-chosen area. Our proposed approach is robust to shading, specular highlights, and cast shadows, enabling selection in real images. As we do not rely on semantic segmentation (different woods or metal should not be selected together), we formulate the problem as a similarity-based grouping problem based on a user-provided image location. In particular, we propose to leverage the unsupervised DINO [Caron et al. 2021] features coupled with a proposed Cross-Similarity Feature Weighting module and an MLP head to extract material similarities in an image. We train our model on a new synthetic image dataset, that we release. We show that our method generalizes well to real-world images. We carefully analyze our model's behavior on varying material properties and lighting. Additionally, we evaluate it against a hand-annotated benchmark of 50 real photographs. We further demonstrate our model on a set of applications, including material editing, in-video selection, and retrieval of object photographs with similar materials.

Generative transformers have experienced rapid popularity growth in the computer vision community in synthesizing high-fidelity and high-resolution images. The best generative transformer models so far, however, still treat an image naively as a sequence of tokens, and decode an image sequentially following the raster scan ordering (i.e. line-by-line). We find this strategy neither optimal nor efficient. This paper proposes a novel image synthesis paradigm using a bidirectional transformer decoder, which we term MaskGIT. During training, MaskGIT learns to predict randomly masked tokens by attending to tokens in all directions. At inference time, the model begins with generating all tokens of an image simultaneously, and then refines the image iteratively conditioned on the previous generation. Our experiments demonstrate that MaskGIT significantly outperforms the state-of-the-art transformer model on the ImageNet dataset, and accelerates autoregressive decoding by up to 48x. Besides, we illustrate that MaskGIT can be easily extended to various image editing tasks, such as inpainting, extrapolation, and image manipulation. Project page: masked-generative-image-transformer.github.io.

We explore a feed-forward approach for decomposing a video into layers, where each layer contains an object of interest along with its associated shadows, reflections, and other visual effects. This problem is challenging since associated effects vary widely with the 3D geometry and lighting conditions in the scene, and ground-truth labels for visual effects are difficult (and in some cases impractical) to collect. We take a self-supervised approach and train a neural network to produce a foreground image and alpha matte from a rough object segmentation mask under a reconstruction and sparsity loss. Under reconstruction loss, the layer decomposition problem is underdetermined: many combinations of layers may reconstruct the input video.Inspired by the game theory concept of focal points---or \emph{Schelling points}---we pose the problem as a coordination game, where each player (network) predicts the effects for a single object without knowledge of the other players' choices. The players learn to converge on the ``natural'' layer decomposition in order to maximize the likelihood of their choices aligning with the other players'. We train the network to play this game with itself, and show how to design the rules of this game so that the focal point lies at the correct layer decomposition. We demonstrate feed-forward results on a challenging synthetic dataset, then show that pretraining on this dataset significantly reduces optimization time for real videos.

Computer vision is increasingly effective at segmenting objects in images and videos; however, scene effects related to the objects—shadows, reflections, generated smoke, etc.—are typically overlooked. Identifying such scene effects and associating them with the objects producing them is important for improving our fundamental understanding of visual scenes, and can also assist a variety of applications such as removing, duplicating, or enhancing objects in video. In this work, we take a step towards solving this novel problem of automatically associating objects with their effects in video. Given an ordinary video and a rough segmentation mask over time of one or more subjects of interest, we estimate an omnimatte for each subject—an alpha matte and color image that includes the subject along with all its related time-varying scene elements. Our model is trained only on the input video in a self-supervised manner, without any manual labels, and is generic—it produces omnimattes automatically for arbitrary objects and a variety of effects. We show results on real-world videos containing interactions between different types of subjects (cars, animals, people) and complex effects, ranging from semitransparent elements such as smoke and reflections, to fully opaque effects such as objects attached to the subject.1

Synthetic datasets play a critical role in pre-training CNN models for optical flow, but they are painstaking to generate and hard to adapt to new applications. To automate the process, we present AutoFlow, a simple and effective method to render training data for optical flow that optimizes the performance of a model on a target dataset. AutoFlow takes a layered approach to render synthetic data, where the motion, shape, and appearance of each layer are controlled by learnable hyperparameters. Experimental results show that AutoFlow achieves state-of-the-art accuracy in pre-training both PWC-Net and RAFT. Our code and data are available at autoflow-google.github.io.

Remarkable progress has been made in 3D reconstruction of rigid structures from a video or a collection of images. However, it is still challenging to reconstruct nonrigid structures from RGB inputs, due to its under-constrained nature. While template-based approaches, such as parametric shape models, have achieved great success in modeling the "closed world" of known object categories, they cannot well handle the "open-world" of novel object categories or outlier shapes. In this work, we introduce a template-free approach to learn 3D shapes from a single video. It adopts an analysis-by-synthesis strategy that forward-renders object silhouette, optical flow, and pixel values to compare with video observations, which generates gradients to adjust the camera, shape and motion parameters. Without using a category-specific shape template, our method faithfully reconstructs nonrigid 3D structures from videos of human, animals, and objects of unknown classes. Our code is available at lasr-google.github.io.

We present deep neural network methodology to reconstruct the 3d pose and shape of people, including hand gestures and facial expression, given an input RGB image. We rely on a recently introduced, expressive full body statistical 3d human model, GHUM, trained end-to-end, and learn to reconstruct its pose and shape state in a self-supervised regime. Central to our methodology, is a learning to learn and optimize approach, referred to as HUman Neural Descent (HUND), which avoids both second-order differentiation when training the model parameters, and expensive state gradient descent in order to accurately minimize a semantic differentiable rendering loss at test time. Instead, we rely on novel recurrent stages to update the pose and shape parameters such that not only losses are minimized effectively, but the process is meta-regularized in order to ensure endprogress. HUND’s symmetry between training and testing makes it the first 3d human sensing architecture to natively support different operating regimes including self-supervised ones. In diverse tests, we show that HUND achieves very competitive results in datasets like H3.6M and 3DPW, as well as good quality 3d reconstructions for complex imagery collected in-the-wild.

Single image 3D photography enables viewers to view a still image from novel viewpoints. Recent approaches combine monocular depth networks with inpainting networks to achieve compelling results. A drawback of these techniques is the use of hard depth layering, making them unable to model intricate appearance details such as thin hair-like structures. We present SLIDE, a modular and unified system for single image 3D photography that uses a simple yet effective soft layering strategy to better preserve appearance details in novel views. In addition, we propose a novel depth-aware training strategy for our inpainting module, better suited for the 3D photography task. The resulting SLIDE approach is modular, enabling the use of other components such as segmentation and matting for improved layering. At the same time, SLIDE uses an efficient layered depth formulation that only requires a single forward pass through the component networks to produce high quality 3D photos. Extensive experimental analysis on three view-synthesis datasets, in combination with user studies on in-the-wild image collections, demonstrate superior performance of our technique in comparison to existing strong baselines while being conceptually much simpler. Project page: https://varunjampani.github.io/slide

Image classification models can depend on multiple different semantic attributes of the image. An explanation of the decision of the classifier needs to both discover and visualize these properties. Here we present StylEx, a method for doing this, by training a generative model to specifically explain multiple attributes that underlie classifier decisions. A natural source for such attributes is the StyleSpace of StyleGAN, which is known to generate semantically meaningful dimensions in the image. However, because standard GAN training is not dependent on the classifier, it may not represent those attributes which are important for the classifier decision, and the dimensions of StyleSpace may represent irrelevant at-tributes. To overcome this, we propose a training procedure for a StyleGAN, which incorporates the classifier model, in order to learn a classifier-specific StyleSpace. Explanatory attributes are then selected from this space. These can be used to visualize the effect of changing multiple attributes per image, thus providing image-specific explanations. We apply StylEx to multiple domains, including animals, leaves, faces and retinal images. For these, we show how an image can be modified in different ways to change its classifier output. Our results show that the method finds attributes that align well with semantic ones, generate meaningful image-specific explanations, and are human-interpretable as measured in user-studies.1

We present a passive non-line-of-sight method that infers the number of people or activity of a person from the observation of a blank wall in an unknown room. Our technique analyzes complex imperceptible changes in indirect illumination in a video of the wall to reveal a signal that is correlated with motion in the hidden part of a scene. We use this signal to classify between zero, one, or two moving people, or the activity of a person in the hidden scene. We train two convolutional neural networks using data collected from 20 different scenes, and achieve an accuracy of ≈ 94% for both tasks in unseen test environments and real-time online settings. Unlike other passive non-line-of-sight methods, the technique does not rely on known occluders or controllable light sources, and generalizes to unknown rooms with no recalibration. We analyze the generalization and robustness of our method with both real and synthetic data, and study the effect of the scene parameters on the signal quality.1

We present THUNDR, a transformer-based deep neural network methodology to reconstruct the 3d pose and shape of people, given monocular RGB images. Key to our methodology is an intermediate 3d marker representation, where we aim to combine the predictive power of model-free-output architectures and the regularizing, anthropometrically-preserving properties of a statistical human surface model like GHUM—a recently introduced, expressive full body statistical 3d human model, trained end-to-end. Our novel transformer-based prediction pipeline can focus on image regions relevant to the task, supports self-supervised regimes, and ensures that solutions are consistent with human anthropometry. We show state-of-the-art results on Human3.6M and 3DPW, for both the fully-supervised and the self-supervised models, for the task of inferring 3d human shape, joint positions, and global translation. Moreover, we observe very solid 3d reconstruction performance for difficult human poses collected in the wild.

Inferring representations of 3D scenes from 2D observations is a fundamental problem of computer graphics, computer vision, and artificial intelligence. Emerging 3D-structured neural scene representations are a promising approach to 3D scene understanding. In this work, we propose a novel neural scene representation, Light Field Networks or LFNs, which represent both geometry and appearance of the underlying 3D scene in a 360-degree, four-dimensional light field parameterized via a neural implicit representation. Rendering a ray from an LFN requires only a single network evaluation, as opposed to hundreds of evaluations per ray for ray-marching or volumetric based renderers in 3D-structured neural scene representations. In the setting of simple scenes, we leverage meta-learning to learn a prior over LFNs that enables multi-view consistent light field reconstruction from as little as a single image observation. This results in dramatic reductions in time and memory complexity, and enables real-time rendering. The cost of storing a 360-degree light field via an LFN is two orders of magnitude lower than conventional methods such as the Lumigraph. Utilizing the analytical differentiability of neural implicit representations and a novel parameterization of light space, we further demonstrate the extraction of sparse depth maps from LFNs.

We present a method for predicting dense depth in scenarios where both a monocular camera and people in the scene are freely moving (right). Existing methods for recovering depth for dynamic, non-rigid objects from monocular video impose strong assumptions on the objects' motion and may only recover sparse depth. In this paper, we take a data-driven approach and learn human depth priors from a new source of data: thousands of Internet videos of people imitating mannequins, i.e., freezing in diverse, natural poses, while a hand-held camera tours the scene (left). Because people are stationary, geometric constraints hold, thus training data can be generated using multi-view stereo reconstruction. At inference time, our method uses motion parallax cues from the static areas of the scenes to guide the depth prediction. We evaluate our method on real-world sequences of complex human actions captured by a moving hand-held camera, show improvement over state-of-the-art monocular depth prediction methods, and demonstrate various 3D effects produced using our predicted depth.

High-resolution simulations can deliver great visual quality, but they are often limited by available memory, especially on GPUs. We present a compiler for physical simulation that can achieve both high performance and significantly reduced memory costs, by enabling flexible and aggressive quantization. Low-precision ("quantized") numerical data types are used and packed to represent simulation states, leading to reduced memory space and bandwidth consumption. Quantized simulation allows higher resolution simulation with less memory, which is especially attractive on GPUs. Implementing a quantized simulator that has high performance and packs the data tightly for aggressive storage reduction would be extremely labor-intensive and error-prone using a traditional programming language. To make the creation of quantized simulation practical, we have developed a new set of language abstractions and a compilation system. A suite of tailored domain-specific optimizations ensure quantized simulators often run as fast as the full-precision simulators, despite the overhead of encoding-decoding the packed quantized data types. Our programming language and compiler, based on Taichi , allow developers to effortlessly switch between different full-precision and quantized simulators, to explore the full design space of quantization schemes, and ultimately to achieve a good balance between space and precision. The creation of quantized simulation with our system has large benefits in terms of memory consumption and performance, on a variety of hardware, from mobile devices to workstations with high-end GPUs. We can simulate with levels of resolution that were previously only achievable on systems with much more memory, such as multiple GPUs. For example, on a single GPU, we can simulate a Game of Life with 20 billion cells (8× compression per pixel), an Eulerian fluid system with 421 million active voxels (1.6× compression per voxel), and a hybrid Eulerian-Lagrangian elastic object simulation with 235 million particles (1.7× compression per particle). At the same time, quantized simulations create physically plausible results. Our quantization techniques are complementary to existing acceleration approaches of physical simulation: they can be used in combination with these existing approaches, such as sparse data structures, for even higher scalability and performance.

We present a method to estimate depth of a dynamic scene, containing arbitrary moving objects, from an ordinary video captured with a moving camera. We seek a geometrically and temporally consistent solution to this under-constrained problem: the depth predictions of corresponding points across frames should induce plausible, smooth motion in 3D. We formulate this objective in a new test-time training framework where a depth-prediction CNN is trained in tandem with an auxiliary scene-flow prediction MLP over the entire input video. By recursively unrolling the scene-flow prediction MLP over varying time steps, we compute both short-range scene flow to impose local smooth motion priors directly in 3D, and long-range scene flow to impose multi-view consistency constraints with wide baselines. We demonstrate accurate and temporally coherent results on a variety of challenging videos containing diverse moving objects (pets, people, cars), as well as camera motion. Our depth maps give rise to a number of depth-and-motion aware video editing effects such as object and lighting insertion.

We address the problem of recovering the shape and spatially-varying reflectance of an object from multi-view images (and their camera poses) of an object illuminated by one unknown lighting condition. This enables the rendering of novel views of the object under arbitrary environment lighting and editing of the object's material properties. The key to our approach, which we call Neural Radiance Factorization (NeRFactor), is to distill the volumetric geometry of a Neural Radiance Field (NeRF) [Mildenhall et al. 2020] representation of the object into a surface representation and then jointly refine the geometry while solving for the spatially-varying reflectance and environment lighting. Specifically, NeRFactor recovers 3D neural fields of surface normals, light visibility, albedo, and Bidirectional Reflectance Distribution Functions (BRDFs) without any supervision, using only a re-rendering loss, simple smoothness priors, and a data-driven BRDF prior learned from real-world BRDF measurements. By explicitly modeling light visibility, NeRFactor is able to separate shadows from albedo and synthesize realistic soft or hard shadows under arbitrary lighting conditions. NeRFactor is able to recover convincing 3D models for free-viewpoint relighting in this challenging and underconstrained capture setup for both synthetic and real scenes. Qualitative and quantitative experiments show that NeRFactor outperforms classic and deep learning-based state of the art across various tasks. Our videos, code, and data are available at people.csail.mit.edu/xiuming/projects/nerfactor/.

Deploying Machine Learning (ML) algorithms within databases is a challenge due to the varied computational footprints of modern ML algorithms and the myriad of database technologies each with their own restrictive syntax. We introduce an Apache Spark-based micro-service orchestration framework that extends database operations to include web service primitives. Our system can orchestrate web services across hundreds of machines and takes full advantage of cluster, thread, and asynchronous parallelism. Using this framework, we provide large scale clients for intelligent services such as speech, vision, search, anomaly detection, and text analysis. This allows users to integrate ready-to-use intelligence into any datastore with an Apache Spark connector. To eliminate the majority of overhead from network communication, we also introduce a low-latency containerized version of our architecture. Finally, we demonstrate that the services we investigate are competitive on a variety of benchmarks, and present two applications of this framework to create intelligent search engines, and real time auto race analytics systems.

We wish to automatically predict the “speediness” of moving objects in videos - whether they move faster, at, or slower than their “natural” speed. The core component in our approach is SpeedNet - a novel deep network trained to detect if a video is playing at normal rate, or if it is sped up. SpeedNet is trained on a large corpus of natural videos in a self-supervised manner, without requiring any manual annotations. We show how this single, binary classification network can be used to detect arbitrary rates of speediness of objects. We demonstrate prediction results by SpeedNet on a wide range of videos containing complex natural motions, and examine the visual cues it utilizes for making those predictions. Importantly, we show that through predicting the speed of videos, the model learns a powerful and meaningful space-time representation that goes beyond simple motion cues. We demonstrate how those learned features can boost the performance of self supervised action recognition, and can be used for video retrieval. Furthermore, we also apply SpeedNet for generating time-varying, adaptive video speedups, which can allow viewers to watch videos faster, but with less of the jittery, unnatural motions typical to videos that are sped up uniformly.

We study the inverse graphics problem of inferring a holistic representation for natural images. Given an input image, our goal is to induce a neuro-symbolic, program-like representation that jointly models camera poses, object locations, and global scene structures. Such high-level, holistic scene representations further facilitate low-level image manipulation tasks such as inpainting. We formulate this problem as jointly finding the camera pose and scene structure that best describe the input image. The benefits of such joint inference are two-fold: scene regularity serves as a new cue for perspective correction, and in turn, correct perspective correction leads to a simplified scene structure, similar to how the correct shape leads to the most regular texture in shape from texture. Our proposed framework, Perspective Plane Program Induction (P3I), combines search-based and gradient-based algorithms to efficiently solve the problem. P3I outperforms a set of baselines on a collection of Internet images, across tasks including camera pose estimation, global structure inference, and down-stream image manipulation tasks.

We present a novel GAN-based model that utilizes the space of deep features learned by a pre-trained classification model. Inspired by classical image pyramid representations, we construct our model as a Semantic Generation Pyramid -- a hierarchical framework which leverages the continuum of semantic information encapsulated in such deep features; this ranges from low level information contained in fine features to high level, semantic information contained in deeper features. More specifically, given a set of features extracted from a reference image, our model generates diverse image samples, each with matching features at each semantic level of the classification model. We demonstrate that our model results in a versatile and flexible framework that can be used in various classic and novel image generation tasks. These include: generating images with a controllable extent of semantic similarity to a reference image, and different manipulation tasks such as semantically-controlled inpainting and compositing; all achieved with the same model, with no further training.

We present a statistical, articulated 3D human shape modeling pipeline, within a fully trainable, modular, deep learning framework. Given high-resolution complete 3D body scans of humans, captured in various poses, together with additional closeups of their head and facial expressions, as well as hand articulation, and given initial, artist designed, gender neutral rigged quad-meshes, we train all model parameters including non-linear shape spaces based on variational auto-encoders, pose-space deformation correctives, skeleton joint center predictors, and blend skinning functions, in a single consistent learning loop. The models are simultaneously trained with all the 3d dynamic scan data (over 60,000 diverse human configurations in our new dataset) in order to capture correlations and ensure consistency of various components. Models support facial expression analysis, as well as body (with detailed hand) shape and pose estimation. We provide fully train-able generic human models of different resolutions- the moderate-resolution GHUM consisting of 10,168 vertices and the low-resolution GHUML(ite) of 3,194 vertices-, run comparisons between them, analyze the impact of different components and illustrate their reconstruction from image data. The models will be available for research.

We consider two important aspects in understanding and editing images: modeling regular, program-like texture or patterns in 2D planes, and 3D posing of these planes in the scene. Unlike prior work on image-based program synthesis, which assumes the image contains a single visible 2D plane, we present Box Program Induction (BPI), which infers a program-like scene representation that simultaneously models repeated structure on multiple 2D planes, the 3D position and orientation of the planes, and camera parameters, all from a single image. Our model assumes a box prior, i.e., that the image captures either an inner view or an outer view of a box in 3D. It uses neural networks to infer visual cues such as vanishing points, wireframe lines to guide a search-based algorithm to find the program that best explains the image. Such a holistic, structured scene representation enables 3D-aware interactive image editing operations such as inpainting missing pixels, changing camera parameters, and extrapolate the image contents.

We present a method for retiming people in an ordinary, natural video --- manipulating and editing the time in which different motions of individuals in the video occur. We can temporally align different motions, change the speed of certain actions (speeding up/slowing down, or entirely "freezing" people), or "erase" selected people from the video altogether. We achieve these effects computationally via a dedicated learning-based layered video representation, where each frame in the video is decomposed into separate RGBA layers, representing the appearance of different people in the video. A key property of our model is that it not only disentangles the direct motions of each person in the input video, but also correlates each person automatically with the scene changes they generate---e.g., shadows, reflections, and motion of loose clothing. The layers can be individually retimed and recombined into a new video, allowing us to achieve realistic, high-quality renderings of retiming effects for real-world videos depicting complex actions and involving multiple individuals, including dancing, trampoline jumping, or group running.

We present a method for predicting dense depth in scenarios where both a monocular camera and people in the scene are freely moving. Existing methods for recovering depth for dynamic, non-rigid objects from monocular video impose strong assumptions on the objects’ motion and may only recover sparse depth. In this paper, we take a data-driven approach and learn human depth priors from a new source of data: thousands of Internet videos of people imitating mannequins, i.e., freezing in diverse, natural poses, while a hand-held camera tours the scene. Since the people are stationary, training data can be created from these videos using multi-view stereo reconstruction. At inference time, our method uses motion parallax cues from the static areas of the scenes, and shows clear improvement over state-of-the-art monocular depth prediction methods. We demonstrate our method on real-world sequences of complex human actions captured by a moving hand-held camera, and show various 3D effects produced using our predicted depth.

How much can we infer about a person’s looks from the way they speak? In this paper, we study the task of reconstructing a facial image of a person from a short audio recording of that person speaking. We design and train a deep neural network to perform this task using millions of natural Internet/Youtube videos of people speaking. During training, our model learns voice-face correlations that allow it to produce images that capture various physical attributes of the speakers such as age, gender and ethnicity. This is done in a self-supervised manner, by utilizing the natural co-occurrence of faces and speech in Internet videos, without the need to model attributes explicitly. We evaluate and numerically quantify how–-and in what manner–-our Speech2Face reconstructions, obtained directly from audio, resemble the true face images of the speakers.

We consider the challenging problem of inferring a hidden moving scene from faint shadows cast on a diffuse surface. Recent work in passive non-line-of-sight (NLoS) imaging has shown that the presence of occluding objects in between the scene and the diffuse surface significantly improves the conditioning of the problem. However, that work assumes that the shape of the occluder is known a priori. In this paper, we relax this often impractical assumption, extending the range of applications for passive occluder-based NLoS imaging systems. We formulate the task of jointly recovering the unknown scene and unknown occluder as a blind deconvolution problem, for which we propose a simple but effective two-step algorithm. At the first step, the algorithm exploits motion in the scene in order to obtain an estimate of the occluder. In particular, it exploits the fact that motion in realistic scenes is typically sparse. The second step is more standard: using regularization, we deconvolve by the occluder estimate to solve for the hidden scene. We demonstrate the effectiveness of our method with simulations and experiments in a variety of settings.

The ability to image scenery outside a camera's line-of-sight would be useful in a variety of applications, including autonomous vehicle collision avoidance, or for first responders to anticipate danger around a corner. When a wall obstructs the camera, light cast onto the floor from behind the wall may be used to recover angular variation of light intensity reflected by the hidden scene, forming a 1D scene projection. Recent work has demonstrated that temporal variation in a video, or sequence of floor images, may be used to image moving components of the hidden scene. However, in many applications, it would be useful to be able to image stationary components as well. This earlier approach was also designed for, and tested on, floors that have approximately uniform albedo, while many real floors have spatially varying albedo patterns such as checkered tiles and patterned carpets. In this work, we propose a method to reconstruct a 1D projection of all components in a hidden scene from a single photograph of the floor without assuming uniform floor albedo. Specifically, we derive a forward model that describes the measured photograph as a nonlinear combination of the unknown floor albedo and the light from behind the wall. The inverse problem, which is the joint estimation of floor albedo and a 1D reconstruction of the hidden scene, is then solved via optimization, where we introduce regularizers that help separate light variations in the measured photograph due to floor pattern and hidden scene, respectively. We demonstrate the effectiveness of our formulation and algorithm using synthetic and experimentally measured data.

We introduce visual deprojection: the task of recovering an image or video that has been collapsed along a dimension. Projections arise in various contexts, such as long-exposure photography, where a dynamic scene is collapsed in time to produce a motion-blurred image, and corner cameras, where reflected light from a scene is collapsed along a spatial dimension because of an edge occluder to yield a 1D video. Deprojection is ill-posed-- often there are many plausible solutions for a given input. We first propose a probabilistic model capturing the ambiguity of the task. We then present a variational inference strategy using convolutional neural networks as functional approximators. Sampling from the inference network at test time yields plausible candidates from the distribution of original signals that are consistent with a given input projection. We evaluate the method on several datasets for both spatial and temporal deprojection tasks. We first demonstrate the method can recover human gait videos and face images from spatial projections, and then show that it can recover videos of moving digits from dramatically motion-blurred images obtained via temporal projection.

Template 3D shapes are useful for many tasks in graphics and vision, including fitting observation data, analyzing shape collections, and transferring shape attributes. Because of the variety of geometry and topology of real-world shapes, previous methods generally use a library of hand-made templates. In this paper, we investigate learning a general shape template from data. To allow for widely varying geometry and topology, we choose an implicit surface representation based on composition of local shape elements. While long known to computer graphics, this representation has not yet been explored in the context of machine learning for vision. We show that structured implicit functions are suitable for learning and allow a network to smoothly and simultaneously fit multiple classes of shapes. The learned shape template supports applications such as shape exploration, correspondence, abstraction, interpolation, and semantic segmentation from an RGB image.

Image extension models have broad applications in image editing, computational photography and computer graphics. While image inpainting has been extensively studied in the literature, it is challenging to directly apply the state-of-the-art inpainting methods to image extension as they tend to generate blurry or repetitive pixels with inconsistent semantics. We introduce semantic conditioning to the discriminator of a generative adversarial network (GAN), and achieve strong results on image extension with coherent semantics and visually pleasing colors and textures. We also show promising results in extreme extensions, such as panorama generation.

Humans are capable of building holistic representations for images at various levels, from local objects, to pairwise relations, to global structures. The interpretation of structures involves reasoning over repetition and symmetry of the objects in the image. In this paper, we present the Program-Guided Image Manipulator (PG-IM), inducing neuro-symbolic program-like representations to represent and manipulate images. Given an image, PG-IM detects repeated patterns, induces symbolic programs, and manipulates the image using a neural network that is guided by the program. PG-IM learns from a single image, exploiting its internal statistics. Despite trained only on image inpainting, PG-IM is directly capable of extrapolation and regularity editing in a unified framework. Extensive experiments show that PG-IM achieves superior performance on all the tasks.

Physical simulators have been widely used in robot planning and control. Among them, differentiable simulators are particularly favored, as they can be incorporated into gradient-based optimization algorithms that are efficient in solving inverse problems such as optimal control and motion planning. Therefore, rigid body simulators and recently their differentiable variants are studied extensively. Simulating deformable objects is, however, more challenging compared to rigid body dynamics. The underlying physical laws of deformable objects are more complex, and the resulting systems have orders of magnitude more degrees of freedom and there-fore they are significantly more computationally expensive to simulate. Computing gradients with respect to physical design or controller parameters is typically even more computationally challenging. In this paper, we propose a real-time, differentiable hybrid Lagrangian-Eulerian physical simulator for deformable objects, ChainQueen, based on the Moving Least Squares Material Point Method (MLS-MPM). MLS-MPM can simulate deformable objects with collisions and can be seamlessly incorporated into soft robotic systems. We demonstrate that our simulator achieves high precision in both forward simulation and backward gradient computation. We have successfully employed it in a diverse set of inference, control and co-design tasks for soft robotics.

We recover a video of the motion taking place in a hidden scene by observing changes in indirect illumination in a nearby uncalibrated visible region. We solve this problem by factoring the observed video into a matrix product between the unknown hidden scene video and an unknown light transport matrix. This task is extremely ill-posed, as any non-negative factorization will satisfy the data. Inspired by recent work on the Deep Image Prior, we parameterize the factor matrices using randomly initialized convolutional neural networks trained in a one-off manner, and show that this results in decompositions that reflect the true motion in the hidden scene.

We study the problem of synthesizing a number of likely future frames from a single input image. In contrast to traditional methods that have tackled this problem in a deterministic or non-parametric way, we propose to model future frames in a probabilistic manner. Our probabilistic model makes it possible for us to sample and synthesize many possible future frames from a single input image. To synthesize realistic movement of objects, we propose a novel network structure, namely a Cross Convolutional Network; this network encodes image and motion information as feature maps and convolutional kernels, respectively. In experiments, our model performs well on synthetic data, such as 2D shapes and animated game sprites, and on real-world video frames. We present analyses of the learned network representations, showing it is implicitly learning a compact encoding of object appearance and motion. We also demonstrate a few of its applications, including visual analogy-making and video extrapolation.

We present an algorithm for creating high resolution anatomically plausible images consistent with acquired clinical brain MRI scans with large inter-slice spacing. Although large data sets of clinical images contain a wealth of information, time constraints during acquisition result in sparse scans that fail to capture much of the anatomy. These characteristics often render computational analysis impractical as many image analysis algorithms tend to fail when applied to such images. Highly specialized algorithms that explicitly handle sparse slice spacing do not generalize well across problem domains. In contrast, we aim to enable application of existing algorithms that were originally developed for high resolution research scans to significantly undersampled scans. We introduce a generative model that captures fine-scale anatomical structure across subjects in clinical image collections and derive an algorithm for filling in the missing data in scans with large inter-slice spacing. Our experimental results demonstrate that the resulting method outperforms state-of-the-art upsampling super-resolution techniques, and promises to facilitate subsequent analysis not previously possible with scans of this quality. Our implementation is freely available at https://github.com/adalca/papago.

Cognitive load has been shown, over hundreds of validated studies, to be an important variable for understanding human performance. However, establishing practical, non-contact approaches for automated estimation of cognitive load under real-world conditions is far from a solved problem. Toward the goal of designing such a system, we propose two novel vision-based methods for cognitive load estimation, and evaluate them on a large-scale dataset collected under real-world driving conditions. Cognitive load is defined by which of 3 levels of a validated reference task the observed subject was performing. On this 3-class problem, our best proposed method of using 3D convolutional neural networks achieves 86.1% accuracy at predicting task-induced cognitive load in a sample of 92 subjects from video alone. This work uses the driving context as a training and evaluation dataset, but the trained network is not constrained to the driving environment as it requires no calibration and makes no assumptions about the subject's visual appearance, activity, head pose, scale, and perspective.

We present a method for inferring a 4D light field of a hidden scene from 2D shadows cast by a known occluder on a diffuse wall. We do this by determining how light naturally reflected off surfaces in the hidden scene interacts with the occluder. By modeling the light transport as a linear system, and incorporating prior knowledge about light field structures, we can invert the system to recover the hidden scene. We demonstrate results of our inference method across simulations and experiments with different types of occluders. For instance, using the shadow cast by a real house plant, we are able to recover low resolution light fields with different levels of texture and parallax complexity. We provide two experimental results: a human subject and two planar elements at different depths.

We study the problem of reconstructing an image from information stored at contour locations. We show that high-quality reconstructions with high fidelity to the source image can be obtained from sparse input, e.g., comprising less than 6% of image pixels. This is a significant improvement over existing contour-based reconstruction methods that require much denser input to capture subtle texture information and to ensure image quality. Our model, based on generative adversarial networks, synthesizes texture and details in regions where no input information is provided. The semantic knowledge encoded into our model and the sparsity of the input allows to use contours as an intuitive interface for semantically-aware image manipulation: local edits in contour domain translate to long-range and coherent changes in pixel space. We can perform complex structural changes such as changing facial expression by simple edits of contours. Our experiments demonstrate that humans as well as a face recognition system mostly cannot distinguish between our reconstructions and the source images.

We present a method for training a regression network from image pixels to 3D morphable model coordinates using only unlabeled photographs. The training loss is based on features from a facial recognition network, computed on-the-fly by rendering the predicted faces with a differentiable renderer. To make training from features feasible and avoid network fooling effects, we introduce three objectives: a batch distribution loss that encourages the output distribution to match the distribution of the morphable model, a loopback loss that ensures the network can correctly reinterpret its own output, and a multi-view identity loss that compares the features of the predicted 3D face and the input photograph from multiple viewing angles. We train a regression network using these objectives, a set of unlabeled photographs, and the morphable model itself, and demonstrate state-of-the-art results.

We study 3D shape modeling from a single image and make contributions to it in three aspects. First, we present Pix3D, a large-scale benchmark of diverse image-shape pairs with pixel-level 2D-3D alignment. Pix3D has wide applications in shape-related tasks including reconstruction, retrieval, viewpoint estimation, etc. Building such a large-scale dataset, however, is highly challenging; existing datasets either contain only synthetic data, or lack precise alignment between 2D images and 3D shapes, or only have a small number of images. Second, we calibrate the evaluation criteria for 3D shape reconstruction through behavioral studies, and use them to objectively and systematically benchmark cutting-edge reconstruction algorithms on Pix3D. Third, we design a novel model that simultaneously performs 3D reconstruction and pose estimation; our multi-task learning approach achieves state-of-the-art performance on both tasks.

We seek to understand the arrow of time in videos - what makes videos look like they are playing forwards or backwards? Can we visualize the cues? Can the arrow of time be a supervisory signal useful for activity analysis? To this end, we build three large-scale video datasets and apply a learning-based approach to these tasks. To learn the arrow of time efficiently and reliably, we design a ConvNet suitable for extended temporal footprints and for class activation visualization, and study the effect of artificial cues, such as cinematographic conventions, on learning. Our trained model achieves state-of-the-art performance on large-scale real-world video datasets. Through cluster analysis and localization of important regions for the prediction, we examine learned visual cues that are consistent among many samples and show when and where they occur. Lastly, we use the trained ConvNet for two applications: self-supervision for action recognition, and video forensics - determining whether Hollywood film clips have been deliberately reversed in time, often used as special effects.

Perceiving accurate 3D object shape is important for robots to interact with the physical world. Current research along this direction has been primarily relying on visual observations. Vision, however useful, has inherent limitations due to occlusions and the 2D-3D ambiguities, especially for perception with a monocular camera. In contrast, touch gets precise local shape information, though its efficiency for reconstructing the entire shape could be low. In this paper, we propose a novel paradigm that efficiently perceives accurate 3D object shape by incorporating visual and tactile observations, as well as prior knowledge of common object shapes learned from large-scale shape repositories. We use vision first, applying neural networks with learned shape priors to predict an object's 3D shape from a single-view color image. We then use tactile sensing to refine the shape; the robot actively touches the object regions where the visual prediction has high uncertainty. Our method efficiently builds the 3D shape of common objects from a color image and a small number of tactile explorations (around 10). Our setup is easy to apply and has potentials to help robots better perform grasping or manipulation tasks on real-world objects.

Human scene understanding uses a variety of visual and non-visual cues to perform inference on object types, poses, and relations. Physics is a rich and universal cue which we exploit to enhance scene understanding. We integrate the physical cue of stability into the learning process using a REINFORCE approach coupled to a physics engine, and apply this to the problem of producing the 3D bounding boxes and poses of objects in a scene. We first show that applying physics supervision to an existing scene understanding model increases performance, produces more stable predictions, and allows training to an equivalent performance level with fewer annotated training examples. We then present a novel architecture for 3D scene parsing named Prim R-CNN, learning to predict bounding boxes as well as their 3D size, translation, and rotation. With physics supervision, Prim R-CNN outperforms existing scene understanding approaches on this problem. Finally, we show that applying physics supervision on unlabeled real images improves real domain transfer of models training on synthetic data.

Deep neural networks, trained with large amount of labeled data, can fail to generalize well when tested with examples from a target domain whose distribution differs from the training data distribution, referred as the source domain. It can be expensive or even infeasible to obtain required amount of labeled data in all possible domains. Unsupervised domain adaptation sets out to address this problem, aiming to learn a good predictive model for the target domain using labeled examples from the source domain but only unlabeled examples from the target domain. Domain alignment approaches this problem by matching the source and target feature distributions, and has been used as a key component in many state-of-the-art domain adaptation methods. However, matching the marginal feature distributions does not guarantee that the corresponding class conditional distributions will be aligned across the two domains. We propose co-regularized domain alignment for unsupervised domain adaptation, which constructs multiple diverse feature spaces and aligns source and target distributions in each of them individually, while encouraging that alignments agree with each other with regard to the class predictions on the unlabeled target examples. The proposed method is generic and can be used to improve any domain adaptation method which uses domain alignment. We instantiate it in the context of a recent state-of-the-art method and observe that it provides significant performance improvements on several domain adaptation benchmarks.

We aim to obtain an interpretable, expressive, and disentangled scene representation that contains comprehensive structural and textural information for each object. Previous scene representations learned by neural networks are often uninterpretable, limited to a single object, or lacking 3D knowledge. In this work, we propose 3D scene de-rendering networks (3D-SDN) to address the above issues by integrating disentangled representations for semantics, geometry, and appearance into a deep generative model. Our scene encoder performs inverse graphics, translating a scene into a structured object-wise representation. Our decoder has two components: a differentiable shape renderer and a neural texture generator. The disentanglement of semantics, geometry, and appearance supports 3D-aware scene manipulation, e.g., rotating and moving objects freely while keeping the consistent shape and texture, and changing the object appearance without affecting its shape. Experiments demonstrate that our editing scheme based on 3D-SDN is superior to its 2D counterpart.

From a single image, humans are able to perceive the full 3D shape of an object by exploiting learned shape priors from everyday life. Contemporary single-image 3D reconstruction algorithms aim to solve this task in a similar fashion, but often end up with priors that are highly biased by training classes. Here we present an algorithm, Generalizable Reconstruction (GenRe), designed to capture more generic, class-agnostic shape priors. We achieve this with an inference network and training procedure that combine 2.5D representations of visible surfaces (depth and silhouette), spherical shape representations of both visible and non-visible surfaces, and 3D voxel-based representations, in a principled manner that exploits the causal structure of how 3D shapes give rise to 2D images. Experiments demonstrate that GenRe performs well on single-view shape reconstruction, and generalizes to diverse novel objects from categories not seen during training.

Recent progress in deep generative models has led to tremendous breakthroughs in image generation. While being able to synthesize photorealistic images, existing models lack an understanding of our underlying 3D world. Different from previous works built on 2D datasets and models, we present a new generative model, Visual Object Networks (VONs), synthesizing natural images of objects with a disentangled 3D representation. Inspired by classic graphics rendering pipelines, we unravel the image formation process into three conditionally independent factors---shape, viewpoint, and texture---and present an end-to-end adversarial learning framework that jointly models 3D shape and 2D texture. Our model first learns to synthesize 3D shapes that are indistinguishable from real shapes. It then renders the object's 2.5D sketches (i.e., silhouette and depth map) from its shape under a sampled viewpoint. Finally, it learns to add realistic textures to these 2.5D sketches to generate realistic images. The VON not only generates images that are more realistic than the state-of-the-art 2D image synthesis methods but also enables many 3D operations such as changing the viewpoint of a generated image, shape and texture editing, linear interpolation in texture and shape space, and transferring appearance across different objects and viewpoints.

We present a system that visualizes complex human motion via 3D motion sculptures-a representation that conveys the 3D structure swept by a human body as it moves through space. Our system computes a motion sculpture from an input video, and then embeds it back into the scene in a 3D-aware fashion. The user may also explore the sculpture directly in 3D or physically print it. Our interactive interface allows users to customize the sculpture design, for example, by selecting materials and lighting conditions. To provide this end-to-end workflow, we introduce an algorithm that estimates a human's 3D geometry over time from a set of 2D images, and develop a 3D-aware image-based rendering approach that inserts the sculpture back into the original video. By automating the process, our system takes motion sculpture creation out of the realm of professional artists, and makes it applicable to a wide range of existing video material. By conveying 3D information to users, motion sculptures reveal space-time motion information that is difficult to perceive with the naked eye, and allow viewers to interpret how different parts of the object interact over time. We validate the effectiveness of motion sculptures with user studies, finding that our visualizations are more informative about motion than existing stroboscopic and space-time visualization methods.

We propose a novel method for template matching in unconstrained environments. Its essence is the Best-Buddies Similarity (BBS), a useful, robust, and parameter-free similarity measure between two sets of points. BBS is based on counting the number of Best-Buddies Pairs (BBPs)-pairs of points in source and target sets that are mutual nearest neighbours, i.e., each point is the nearest neighbour of the other. BBS has several key features that make it robust against complex geometric deformations and high levels of outliers, such as those arising from background clutter and occlusions. We study these properties, provide a statistical analysis that justifies them, and demonstrate the consistent success of BBS on a challenging real-world dataset while using different types of features.

We present a joint audio-visual model for isolating a single speech signal from a mixture of sounds such as other speakers and background noise. Solving this task using only audio as input is extremely challenging and does not provide an association of the separated speech signals with speakers in the video. In this paper, we present a deep network-based model that incorporates both visual and auditory signals to solve this task. The visual features are used to "focus" the audio on desired speakers in a scene and to improve the speech separation quality. To train our joint audio-visual model, we introduce AVS peech , a new dataset comprised of thousands of hours of video segments from the Web. We demonstrate the applicability of our method to classic speech separation tasks, as well as real-world scenarios involving heated interviews, noisy bars, and screaming children, only requiring the user to specify the face of the person in the video whose speech they want to isolate. Our method shows clear advantage over state-of-the-art audio-only speech separation in cases of mixed speech. In addition, our model, which is speaker-independent (trained once, applicable to any speaker), produces better results than recent audio-visual speech separation methods that are speaker-dependent (require training a separate model for each speaker of interest).

We present a method for synthesizing a frontal, neutral-expression image of a person's face given an input face photograph. This is achieved by learning to generate facial landmarks and textures from features extracted from a facial-recognition network. Unlike previous generative approaches, our encoding feature vector is largely invariant to lighting, pose, and facial expression. Exploiting this invariance, we train our decoder network using only frontal, neutral-expression photographs. Since these photographs are well aligned, we can decompose them into a sparse set of landmark points and aligned texture maps. The decoder then predicts landmarks and textures independently and combines them using a differentiable image warping operation. The resulting images can be usedfor a number of applications, such as analyzing facial attributes, exposure and white balance adjustment, or creating a 3-D avatar.

Visible watermarking is a widely-used technique for marking and protecting copyrights of many millions of images on the web, yet it suffers from an inherent security flaw-watermarks are typically added in a consistent manner to many images. We show that this consistency allows to automatically estimate the watermark and recover the original images with high accuracy. Specifically, we present a generalized multi-image matting algorithm that takes a watermarked image collection as input and automatically estimates the foreground (watermark), its alpha matte, and the background (original) images. Since such an attack relies on the consistency of watermarks across image collection, we explore and evaluate how it is affected by various types of inconsistencies in the watermark embedding that could potentially be used to make watermarking more secured. We demonstrate the algorithm on stock imagery available on the web, and provide extensive quantitative analysis on synthetic watermarked data. A key takeaway message of this paper is that visible watermarks should be designed to not only be robust against removal from a single image, but to be more resistant to mass-scale removal from image collections as well.

We show that walls, and other obstructions with edges, can be exploited as naturally-occurring “cameras” that reveal the hidden scenes beyond them. In particular, we demonstrate methods for using the subtle spatio-temporal radiance variations that arise on the ground at the base of a wall's edge to construct a one-dimensional video of the hidden scene behind the wall. The resulting technique can be used for a variety of applications in diverse physical settings. From standard RGB video recordings, we use edge cameras to recover 1-D videos that reveal the number and trajectories of people moving in an occluded scene. We further show that adjacent wall edges, such as those that arise in the case of an open doorway, yield a stereo camera from which the 2-D location of hidden, moving objects can be recovered. We demonstrate our technique in a number of indoor and outdoor environments involving varied floor surfaces and illumination conditions.

Humans infer rich knowledge of objects from both auditory and visual cues. Building a machine of such competency, however, is very challenging, due to the great difficulty in capturing large-scale, clean data of objects with both their appearance and the sound they make. In this paper, we present a novel, open-source pipeline that generates audiovisual data, purely from 3D object shapes and their physical properties. Through comparison with audio recordings and human behavioral studies, we validate the accuracy of the sounds it generates. Using this generative model, we are able to construct a synthetic audio-visual dataset, namely Sound-20K, for object perception tasks. We demonstrate that auditory and visual information play complementary roles in object perception, and further, that the representation learned on synthetic audio-visual data can transfer to real-world scenarios.

We introduce a paradigm for understanding physical scenes without human annotations. At the core of our system is a physical world representation that is first recovered by a perception module and then utilized by physics and graphics engines. During training, the perception module and the generative models learn by visual de-animation --- interpreting and reconstructing the visual information stream. During testing, the system first recovers the physical world state, and then uses the generative models for reasoning and future prediction. Even more so than forward simulation, inverting a physics or graphics engine is a computationally hard problem; we overcome this challenge by using a convolutional inversion network. Our system quickly recognizes the physical world state from appearance and motion cues, and has the flexibility to incorporate both differentiable and non-differentiable physics and graphics engines. We evaluate our system on both synthetic and real datasets involving multiple physical scenes, and demonstrate that our system performs well on both physical state estimation and reasoning problems. We further show that the knowledge learned on the synthetic dataset generalizes to constrained real images.

3D object reconstruction from a single image is a highly under-determined problem, requiring strong prior knowledge of plausible 3D shapes. This introduces challenge for learning-based approaches, as 3D object annotations in real images are scarce. Previous work chose to train on synthetic data with ground truth 3D information, but suffered from the domain adaptation issue when tested on real data. In this work, we propose an end-to-end trainable framework, sequentially estimating 2.5D sketches and 3D object shapes. Our disentangled, two-step formulation has three advantages. First, compared to full 3D shape, 2.5D sketches are much easier to be recovered from a 2D image, and to transfer from synthetic to real data. Second, for 3D reconstruction from the 2.5D sketches, we can easily transfer the learned model on synthetic data to real images, as rendered 2.5D sketches are invariant to object appearance variations in real images, including lighting, texture, etc. This further relieves the domain adaptation problem. Third, we derive differentiable projective functions from 3D shape to 2.5D sketches, making the framework end-to-end trainable on real images, requiring no real-image annotations. Our framework achieves state-of-the-art performance on 3D shape reconstruction.

Hearing an object falling onto the ground, humans can recover rich information including its rough shape, material, and falling height. In this paper, we build machines to approximate such competency. We first mimic human knowledge of the physical world by building an efficient, physics-based simulation engine. Then, we present an analysis-by-synthesis approach to infer properties of the falling object. We further accelerate the process by learning a mapping from a sound wave to object properties, and using the predicted values to initialize the inference. This mapping can be viewed as an approximation of human commonsense learned from past experience. Our model performs well on both synthetic audio clips and real recordings without requiring any annotated data. We conduct behavior studies to compare human responses with ours on estimating object shape, material, and falling height from sound. Our model achieves near-human performance.

The estimation of material properties is important for scene understanding, with many applications in vision, robotics, and structural engineering. This paper connects fundamentals of vibration mechanics with computer vision techniques in order to infer material properties from small, often imperceptible motions in video. Objects tend to vibrate in a set of preferred modes. The frequencies of these modes depend on the structure and material properties of an object. We show that by extracting these frequencies from video of a vibrating object, we can often make inferences about that object's material properties. We demonstrate our approach by estimating material properties for a variety of objects by observing their motion in high-speed and regular frame rate video.

Stereoscopic 3D (S3D) movies have become widely popular in the movie theaters, but the adoption of S3D at home is low even though most TV sets support S3D. It is widely believed that S3D with glasses is not the right approach for the home. A much more appealing approach is to use automulti-scopic displays that provide a glasses-free 3D experience to multiple viewers. A technical challenge is the lack of native multiview content that is required to deliver a proper view of the scene for every viewpoint. Our approach takes advantage of the abundance of stereoscopic 3D movies. We propose a real-time system that can convert stereoscopic video to a high-quality multiview video that can be directly fed to automultiscopic displays. Our algorithm uses a wavelet-based decomposition of stereoscopic images with per-wavelet disparity estimation. A key to our solution lies in combining Lagrangian and Eulerian approaches for both the disparity estimation and novel view synthesis, which leverages the complementary advantages of both techniques. The solution preserves all the features of Eulerian methods, e.g., subpixel accuracy, high performance, robustness to ambiguous depth cases, and easy integration of inter-view aliasing while maintaining the advantages of Lagrangian approaches, e.g., robustness to large disparities and possibility of performing non-trivial disparity manipulations through both view extrapolation and interpolation. The method achieves real-time performance on current GPUs. Its design also enables an easy hardware implementation that is demonstrated using a field-programmable gate array. We analyze the visual quality and robustness of our technique on a number of synthetic and real-world examples. We also perform a user experiment which demonstrates benefits of the technique when compared to existing solutions.

Very long baseline interferometry (VLBI) is a technique for imaging celestial radio emissions by simultaneously observing a source from telescopes distributed across Earth. The challenges in reconstructing images from fine angular resolution VLBI data are immense. The data is extremely sparse and noisy, thus requiring statistical image models such as those designed in the computer vision community. In this paper we present a novel Bayesian approach for VLBI image reconstruction. While other methods often require careful tuning and parameter selection for different types of data, our method (CHIRP) produces good results under different settings such as low SNR or extended emission. The success of our method is demonstrated on realistic synthetic experiments as well as publicly available real data. We present this problem in a way that is accessible to members of the community, and provide a dataset website (vlbiimaging.csail.mit.edu) that facilitates controlled comparisons across algorithms.

Objects make distinctive sounds when they are hit or scratched. These sounds reveal aspects of an object's material properties, as well as the actions that produced them. In this paper, we propose the task of predicting what sound an object makes when struck as a way of studying physical interactions within a visual scene. We present an algorithm that synthesizes sound from silent videos of people hitting and scratching objects with a drumstick. This algorithm uses a recurrent neural network to predict sound features from videos and then produces a waveform from these features with an example-based synthesis procedure. We show that the sounds predicted by our model are realistic enough to fool participants in a "real or fake" psychophysical experiment, and that they convey significant information about material properties and physical interactions.

We study the problem of 3D object generation. We propose a novel framework, namely 3D Generative Adversarial Network (3D-GAN), which generates 3D objects from a probabilistic space by leveraging recent advances in volumetric convolutional networks and generative adversarial nets. The benefits of our model are three-fold: first, the use of an adversarial criterion, instead of traditional heuristic criteria, enables the generator to capture object structure implicitly and to synthesize high-quality 3D objects; second, the generator establishes a mapping from a low-dimensional probabilistic space to the space of 3D objects, so that we can sample objects without a reference image or CAD models, and explore the 3D object manifold; third, the adversarial discriminator provides a powerful 3D shape descriptor which, learned without supervision, has wide applications in 3D object recognition. Experiments demonstrate that our method generates high-quality 3D objects, and our unsupervisedly learned features achieve impressive performance on 3D object recognition, comparable with those of supervised learning methods.

We study the problem of synthesizing a number of likely future frames from a single input image. In contrast to traditional methods, which have tackled this problem in a deterministic or non-parametric way, we propose a novel approach which models future frames in a probabilistic manner. Our proposed method is therefore able to synthesize multiple possible next frames using the same model. Solving this challenging problem involves low- and high-level image and motion understanding for successful image synthesis. Here, we propose a novel network structure, namely a Cross Convolutional Network, that encodes images as feature maps and motion information as convolutional kernels to aid in synthesizing future frames. In experiments, our model performs well on both synthetic data, such as 2D shapes and animated game sprites, as well as on real-wold video data. We show that our model can also be applied to tasks such as visual analogy-making, and present analysis of the learned network representations.

We present a principled framework for inferring pixel labels in weakly-annotated image datasets. Most previous, example-based approaches to computer vision rely on a large corpus of densely labeled images. However, for large, modern image datasets, such labels are expensive to obtain and are often unavailable. We establish a large-scale graphical model spanning all labeled and unlabeled images, then solve it to infer pixel labels jointly for all images in the dataset while enforcing consistent annotations over similar visual patterns. This model requires significantly less labeled data and assists in resolving ambiguities by propagating inferred annotations from images with stronger local visual evidences to images with weaker local evidences. We apply our proposed framework to two computer vision problems, namely image annotation with semantic segmentation, and object discovery and co-segmentation (segmenting multiple images containing a common object). Extensive numerical evaluations and comparisons show that our method consistently outperforms the state-of-the-art in automatic annotation and semantic labeling, while requiring significantly less labeled data. In contrast to previous co-segmentation techniques, our method manages to discover and segment objects well even in the presence of substantial amounts of noise images (images not containing the common object), as typical for datasets collected from Internet search.

This special section contains selected papers from the IEEE Computer Vision and Pattern Recognition (CVPR), June, 2013, jointly sponsored by the IEEE and the Computer Vision Foundation.

The estimation of material properties is important for scene understanding, with many applications in vision, robotics, and structural engineering. This paper connects fundamentals of vibration mechanics with computer vision techniques in order to infer material properties from small, often imperceptible motion in video. Objects tend to vibrate in a set of preferred modes. The shapes and frequencies of these modes depend on the structure and material properties of an object. Focusing on the case where geometry is known or fixed, we show how information about an object's modes of vibration can be extracted from video and used to make inferences about that object's material properties. We demonstrate our approach by estimating material properties for a variety of rods and fabrics by passively observing their motion in high-speed and regular framerate video.

We propose a novel method for template matching in unconstrained environments. Its essence is the Best-Buddies Similarity (BBS), a useful, robust, and parameter-free similarity measure between two sets of points. BBS is based on counting the number of Best-Buddies Pairs (BBPs)-pairs of points in source and target sets, where each point is the nearest neighbor of the other. BBS has several key features that make it robust against complex geometric deformations and high levels of outliers, such as those arising from background clutter and occlusions. We study these properties, provide a statistical analysis that justifies them, and demonstrate the consistent success of BBS on a challenging real-world dataset.

Video magnification reveals subtle variations that would be otherwise invisible to the naked eye. Current techniques require all motion in the video to be very small, which is unfortunately not always the case. Tiny yet meaningful motions are often combined with larger motions, such as the small vibrations of a gate as it rotates, or the microsaccades in a moving eye. We present a layer-based video magnification approach that can amplify small motions within large ones. An examined region/layer is temporally aligned and subtle variations are magnified. Matting is used to magnify only region of interest while maintaining integrity of nearby sites. Results show handling larger motions, larger amplification factors and significant reduction in artifacts over state of the art.

Photographs taken through glass windows often contain both the desired scene and undesired reflections. Separating the reflection and transmission layers is an important but ill-posed problem that has both aesthetic and practical applications. In this work, we introduce the use of ghosting cues that exploit asymmetry between the layers, thereby helping to reduce the ill-posedness of the problem. These cues arise from shifted double reflections of the reflected scene off the glass surface. In double-pane windows, each pane reflects shifted and attenuated versions of objects on the same side of the glass as the camera. For single-pane windows, ghosting cues arise from shifted reflections on the two surfaces of the glass pane. Even though the ghosting is sometimes barely perceptible by humans, we can still exploit the cue for layer separation. In this work, we model the ghosted reflection using a double-impulse convolution kernel, and automatically estimate the spatial separation and relative attenuation of the ghosted reflection components. To separate the layers, we propose an algorithm that uses a Gaussian Mixture Model for regularization. Our method is automatic and requires only a single input image. We demonstrate that our approach removes a large fraction of reflections on both synthetic and real-world inputs.

When viewed through a small aperture, a moving image provides incomplete information about the local motion. Only the component of motion along the local image gradient is constrained. In an essential part of optical flow algorithms, information must be aggregated from nearby image locations in order to estimate all components of motion. This limitation of local evidence for estimating optical flow is called “the aperture problem”. We pose and solve a generalization of the aperture problem for moving refractive elements. We consider a common setup in air flow imaging or telescope observation: a camera is viewing a static background, and an unknown refractive elements undergoing unknown motion between them. Then we are addressing this fundamental question: what does the local image motion tell us about the motion of refractive elements? We show that the information gleaned through a local aperture for this case is very different than that for optical flow. In optical flow, the movement of 1D structure already constrains the motion in a certain direction. However, we cannot infer any information about the refractive motion from the movement of 1D structure in the observed sequence, and can only recover one component of the motion from 2D structure. Results on both simulated and real sequences are shown to illustrate our theory.

We propose a framework that infers mid-level visual properties of an image by learning about ordinal relationships. Instead of estimating metric quantities directly, the system proposes pairwise relationship estimates for points in the input image. These sparse probabilistic ordinal measurements are globalized to create a dense output map of continuous metric measurements. Estimating order relationships between pairs of points has several advantages over metric estimation: it solves a simpler problem than metric regression, humans are better at relative judgements, so data collection is easier, ordinal relationships are invariant to monotonic transformations of the data, thereby increasing the robustness of the system and providing qualitatively different information. We demonstrate that this frame-work works well on two important mid-level vision tasks: intrinsic image decomposition and depth from an RGB image. We train two systems with the same architecture on data from these two modalities. We provide an analysis of the resulting models, showing that they learn a number of simple rules to make ordinal decisions. We apply our algorithm to depth estimation, with good results, and intrinsic image decomposition, with state-of-the-art results.

Humans demonstrate remarkable abilities to predict physical events in dynamic scenes, and to infer the physical properties of objects from static images. We propose a generative model for solving these problems of physical scene understanding from real-world videos and images. At the core of our generative model is a 3D physics engine, operating on an object-based representation of physical properties, including mass, position, 3D shape, and friction. We can infer these latent properties using relatively brief runs of MCMC, which drive simulations in the physics engine to fit key features of visual observations. We further explore directly mapping visual inputs to physical properties, inverting a part of the generative process using deep learning. We name our model Galileo, and evaluate it on a video dataset with simple yet physically rich scenarios. Results show that Galileo is able to infer the physical properties of objects and predict the outcome of a variety of physical events, with an accuracy comparable to human subjects. Our study points towards an account of human vision with generative physical knowledge at its core, and various recognition models as helpers leading to efficient inference.

We present an algorithm for automatically detecting and visualizing small non-local variations between repeating structures in a single image. Our method allows to automatically correct these variations, thus producing an 'idealized' version of the image in which the resemblance between recurring structures is stronger. Alternatively, it can be used to magnify these variations, thus producing an exaggerated image which highlights the various variations that are difficult to spot in the input image. We formulate the estimation of deviations from perfect recurrence as a general optimization problem, and demonstrate it in the particular cases of geometric deformations and color variations.

We present a unified computational approach for taking photos through reflecting or occluding elements such as windows and fences. Rather than capturing a single image, we instruct the user to take a short image sequence while slightly moving the camera. Differences that often exist in the relative position of the background and the obstructing elements from the camera allow us to separate them based on their motions, and to recover the desired background scene as if the visual obstructions were not there. We show results on controlled experiments and many real and practical scenarios, including shooting through reflections, fences, and raindrop-covered windows.

Structures and objects are often supposed to have idealized geometries such as straight lines or circles. Although not always visible to the naked eye, in reality, these objects deviate from their idealized models. Our goal is to reveal and visualize such subtle geometric deviations, which can contain useful, surprising information about our world. Our framework, termed Deviation Magnification , takes a still image as input, fits parametric models to objects of interest, computes the geometric deviations, and renders an output image in which the departures from ideal geometries are exaggerated. We demonstrate the correctness and usefulness of our method through quantitative evaluation on a synthetic dataset and by application to challenging natural images.

Data-driven techniques can reliably build semantic correspondence among images. In this paper, we present a new regularization model for stereo or flow through transferring the shape information of the disparity or flow from semantically matched patches in the training database. Compared to previous regularization models based on image appearance alone, we can better resolve local ambiguity of the disparity or flow by considering the semantic information without explicit object modeling. We incorporate this data-driven regularization model into a standard Markov Random Field (MRF) model, inferred with a gradient descent algorithm and learned with a discriminative learning approach. Compared to prior state-of-the-art methods, our full model achieves comparable or better results on the KITTI stereo and flow datasets, and improves results on the Sintel Flow dataset under an online estimation setting.

Learning a low-dimensional representation of images is useful for various applications in graphics and computer vision. Existing solutions either require manually specified landmarks for corresponding points in the images, or are restricted to specific objects or shape deformations. This paper alleviates these limitations by imposing a specific model for generating images, the nested composition of color, shape, and appearance. We show that each component can be approximated by a low-dimensional subspace when the others are factored out. Our formulation allows for efficient learning and experiments show encouraging results.

We address the problem of camouflaging a 3D object from the many viewpoints that one might see it from. Given photographs of an object's surroundings, we produce a surface texture that will make the object difficult for a human to detect. To do this, we introduce several background matching algorithms that attempt to make the object look like whatever is behind it. Of course, it is impossible to exactly match the background from every possible viewpoint. Thus our models are forced to make trade-offs between different perceptual factors, such as the conspicuousness of the occlusion boundaries and the amount of texture distortion. We use experiments with human subjects to evaluate the effectiveness of these models for the task of camouflaging a cube, finding that they significantly outperform naïve strategies.

We explore whether we can observe Time's Arrow in a temporal sequence - is it possible to tell whether a video is running forwards or backwards? We investigate this somewhat philosophical question using computer vision and machine learning techniques. We explore three methods by which we might detect Time's Arrow in video sequences, based on distinct ways in which motion in video sequences might be asymmetric in time. We demonstrate good video forwards/backwards classification results on a selection of YouTube video clips, and on natively-captured sequences (with no temporally-dependent video compression), and examine what motions the models have learned that help discriminate forwards from backwards time.

Digital color cameras make sub-sampled measurements of color at alternating pixel locations, and then “demosaick” these measurements to create full color images by up-sampling. This allows traditional cameras with restricted processing hardware to produce color images from a single shot, but it requires blocking a majority of the incident light and is prone to aliasing artifacts. In this paper, we introduce a computational approach to color photography, where the sampling pattern and reconstruction process are co-designed to enhance sharpness and photographic speed. The pattern is made predominantly panchromatic, thus avoiding excessive loss of light and aliasing of high spatial-frequency intensity variations. Color is sampled at a very sparse set of locations and then propagated throughout the image with guidance from the un-aliased luminance channel. Experimental results show that this approach often leads to significant reductions in noise and aliasing artifacts, especially in low-light conditions.

We present a new compact image pyramid representation, the Riesz pyramid, that can be used for real-time phase-based motion magnification. Our new representation is less overcomplete than even the smallest two orientation, octave-bandwidth complex steerable pyramid, and can be implemented using compact, efficient linear filters in the spatial domain. Motion-magnified videos produced with this new representation are of comparable quality to those produced with the complex steerable pyramid. When used with phase-based video magnification, the Riesz pyramid phase-shifts image features along only their dominant orientation rather than every orientation like the complex steerable pyramid.

We identify and study two types of “accidental” images that can be formed in scenes. The first is an accidental pinhole camera image. The second class of accidental images are “inverse” pinhole camera images, formed by subtracting an image with a small occluder present from a reference image without the occluder. Both types of accidental cameras happen in a variety of different situations. For example, an indoor scene illuminated by natural light, a street with a person walking under the shadow of a building, etc. The images produced by accidental cameras are often mistaken for shadows or interreflections. However, accidental images can reveal information about the scene outside the image, the lighting conditions, or the aperture by which light enters the scene.

Headshot portraits are a popular subject in photography but to achieve a compelling visual style requires advanced skills that a casual photographer will not have. Further, algorithms that automate or assist the stylization of generic photographs do not perform well on headshots due to the feature-specific, local retouching that a professional photographer typically applies to generate such portraits. We introduce a technique to transfer the style of an example headshot photo onto a new one. This can allow one to easily reproduce the look of renowned artists. At the core of our approach is a new multiscale technique to robustly transfer the local statistics of an example portrait onto a new one. This technique matches properties such as the local contrast and the overall lighting direction while being tolerant to the unavoidable differences between the faces of two different people. Additionally, because artists sometimes produce entire headshot collections in a common style, we show how to automatically find a good example to use as a reference for a given portrait, enabling style transfer without the user having to search for a suitable example for each input. We demonstrate our approach on data taken in a controlled environment as well as on a large set of photos downloaded from the Internet. We show that we can successfully handle styles by a variety of different artists.

When sound hits an object, it causes small vibrations of the object's surface. We show how, using only high-speed video of the object, we can extract those minute vibrations and partially recover the sound that produced them, allowing us to turn everyday objects---a glass of water, a potted plant, a box of tissues, or a bag of chips---into visual microphones. We recover sounds from high-speed footage of a variety of objects with different properties, and use both real and simulated data to examine some of the factors that affect our ability to visually recover sound. We evaluate the quality of recovered sounds using intelligibility and SNR metrics and provide input and recovered audio samples for direct comparison. We also explore how to leverage the rolling shutter in regular consumer cameras to recover audio from standard frame-rate videos, and use the spatial resolution of our method to visualize how sound-related vibrations vary over an object's surface, which we can use to recover the vibration modes of an object.

Passively estimating the intrinsic material properties of deformable objects moving in a natural environment is essential for scene understanding. We present a framework to automatically analyze videos of fabrics moving under various unknown wind forces, and recover two key material properties of the fabric: stiffness and area weight. We extend features previously developed to compactly represent static image textures to describe video textures, such as fabric motion. A discriminatively trained regression model is then used to predict the physical properties of fabric from these features. The success of our model is demonstrated on a new, publicly available database of fabric videos with corresponding measured ground truth material properties. We show that our predictions are well correlated with ground truth measurements of stiffness and density for the fabrics. Our contributions include: (a) a database that can be used for training and testing algorithms for passively predicting fabric properties from video, (b) an algorithm for predicting the material properties of fabric from a video, and (c) a perceptual study of humans' ability to estimate the material properties of fabric from videos and images.

Latent variables models have been applied to a number of computer vision problems. However, the complexity of the latent space is typically left as a free design choice. A larger latent space results in a more expressive model, but such models are prone to over fitting and are slower to perform inference with. The goal of this paper is to regularize the complexity of the latent space and learn which hidden states are really relevant for prediction. Specifically, we propose using group-sparsity-inducing regularizers such as ℓ1-ℓ2to estimate the parameters of Structured SVMs with unstructured latent variables. Our experiments on digit recognition and object detection show that our approach is indeed able to control the complexity of latent space without any significant loss in accuracy of the learnt model.

We present a data-driven method for building dense 3D reconstructions using a combination of recognition and multi-view cues. Our approach is based on the idea that there are image patches that are so distinctive that we can accurately estimate their latent 3D shapes solely using recognition. We call these patches shape anchors, and we use them as the basis of a multi-view reconstruction system that transfers dense, complex geometry between scenes. We "anchor" our 3D interpretation from these patches, using them to predict geometry for parts of the scene that are relatively ambiguous. The resulting algorithm produces dense reconstructions from stereo point clouds that are sparse and noisy, and we demonstrate it on a challenging dataset of real-world, indoor scenes.

Multi-view autostereoscopic displays provide an immersive, glasses-free 3D viewing experience, but they require correctly filtered content from multiple viewpoints. This, however, cannot be easily obtained with current stereoscopic production pipelines. We provide a practical solution that takes a stereoscopic video as an input and converts it to multi-view and filtered video streams that can be used to drive multi-view autostereoscopic displays. The method combines a phase-based video magnification and an interperspective antialiasing into a single filtering process. The whole algorithm is simple and can be efficiently implemented on current GPUs to yield a near real-time performance. Furthermore, the ability to retarget disparity is naturally supported. Our method is robust and works well for challenging video scenes with defocus blur, motion blur, transparent materials, and specularities. We show that our results are superior when compared to the state-of-the-art depth-based rendering methods. Finally, we showcase the method in the context of a real-time 3D videoconferencing system that requires only two cameras.

Recent attempts to fabricate surfaces with custom reflectance functions boast impressive angular resolution, yet their spatial resolution is limited. In this paper we present a method to construct spatially varying reflectance at a high resolution of up to 220dpi, orders of magnitude greater than previous attempts, albeit with a lower angular resolution. The resolution of previous approaches is limited by the machining, but more fundamentally, by the geometric optics model on which they are built. Beyond a certain scale geometric optics models break down and wave effects must be taken into account. We present an analysis of incoherent reflectance based on wave optics and gain important insights into reflectance design. We further suggest and demonstrate a practical method, which takes into account the limitations of existing micro-fabrication techniques such as photolithography to design and fabricate a range of reflection effects, based on wave interference.

We introduce "time hallucination": synthesizing a plausible image at a different time of day from an input image. This challenging task often requires dramatically altering the color appearance of the picture. In this paper, we introduce the first data-driven approach to automatically creating a plausible-looking photo that appears as though it were taken at a different time of day. The time of day is specified by a semantic time label, such as "night". Our approach relies on a database of time-lapse videos of various scenes. These videos provide rich information about the variations in color appearance of a scene throughout the day. Our method transfers the color appearance from videos with a similar scene as the input photo. We propose a locally affine model learned from the video for the transfer, allowing our model to synthesize new color data while retaining image details. We show that this model can hallucinate a wide range of different times of day. The model generates a large sparse linear system, which can be solved by off-the-shelf solvers. We validate our methods by synthesizing transforming photos of various outdoor scenes to four times of interest: daytime, the golden hour, the blue hour, and nighttime.

We introduce a technique to manipulate small movements in videos based on an analysis of motion in complex-valued image pyramids. Phase variations of the coefficients of a complex-valued steerable pyramid over time correspond to motion, and can be temporally processed and amplified to reveal imperceptible motions, or attenuated to remove distracting changes. This processing does not involve the computation of optical flow, and in comparison to the previous Eulerian Video Magnification method it supports larger amplification factors and is significantly less sensitive to noise. These improved capabilities broaden the set of applications for motion processing in videos. We demonstrate the advantages of this approach on synthetic and natural video sequences, and explore applications in scientific analysis, visualization and video enhancement.

It is often desirable to detect whether a surface has been touched, even when the changes made to that surface are too subtle to see in a pair of before and after images. To address this challenge, we introduce a new imaging technique that combines computational photography and laser speckle imaging. Without requiring controlled laboratory conditions, our method is able to detect surface changes that would be indistinguishable in regular photographs. It is also mobile and does not need to be present at the time of contact with the surface, making it well suited for applications where the surface of interest cannot be constantly monitored. Our approach takes advantage of the fact that tiny surface deformations cause phase changes in reflected coherent light which alter the speckle pattern visible under laser illumination. We take before and after images of the surface under laser light and can detect subtle contact by correlating the speckle patterns in these images. A key challenge we address is that speckle imaging is very sensitive to the location of the camera, so removing and reintroducing the camera requires high-accuracy viewpoint alignment. To this end, we use a combination of computational rephotography and correlation analysis of the speckle pattern as a function of camera translation. Our technique provides a reliable way of detecting subtle surface contact at a level that was previously only possible under laboratory conditions. With our system, the detection of these subtle surface changes can now be brought into the wild.

We identify and study two types of “accidental” images that can be formed in scenes. The first is an accidental pinhole camera image. These images are often mistaken for shadows, but can reveal structures outside a room, or the unseen shape of the light aperture into the room. The second class of accidental images are “inverse” pinhole camera images, formed by subtracting an image with a small occluder present from a reference image without the occluder. The reference image can be an earlier frame of a video sequence. Both types of accidental images happen in a variety of different situations (an indoor scene illuminated by natural light, a street with a person walking under the shadow of a building, etc.). Accidental cameras can reveal information about the scene outside the image, the lighting conditions, or the aperture by which light enters the scene.

The restoration of a blurry or noisy image is commonly performed with a MAP estimator, which maximizes a posterior probability to reconstruct a clean image from a degraded image. A MAP estimator, when used with a sparse gradient image prior, reconstructs piecewise smooth images and typically removes textures that are important for visual realism. We present an alternative deconvolution method called iterative distribution reweighting (IDR) which imposes a global constraint on gradients so that a reconstructed image should have a gradient distribution similar to a reference distribution. In natural images, a reference distribution not only varies from one image to another, but also within an image depending on texture. We estimate a reference distribution directly from an input image for each texture segment. Our algorithm is able to restore rich mid-frequency textures. A large-scale user study supports the conclusion that our algorithm improves the visual realism of reconstructed images compared to those of MAP estimators.

Our goal is to reveal temporal variations in videos that are difficult or impossible to see with the naked eye and display them in an indicative manner. Our method, which we call Eulerian Video Magnification, takes a standard video sequence as input, and applies spatial decomposition, followed by temporal filtering to the frames. The resulting signal is then amplified to reveal hidden information. Using our method, we are able to visualize the flow of blood as it fills the face and also to amplify and reveal small motions. Our technique can run in real time to show phenomena occurring at the temporal frequencies selected by the user.

Camera shake is a common source of degradation in photographs. Restoring blurred pictures is challenging because both the blur kernel and the sharp image are unknown, which makes this problem severely underconstrained. In this work, we estimate camera shake by analyzing edges in the image, effectively constructing the Radon transform of the kernel. Building upon this result, we describe two algorithms for estimating spatially invariant blur kernels. In the first method, we directly invert the transform, which is computationally efficient since it is not necessary to also estimate the latent sharp image. This approach is well suited for scenes with a diversity of edges, such as man-made environments. In the second method, we incorporate the Radon transform within the MAP estimation framework to jointly estimate the kernel and the image. While more expensive, this algorithm performs well on a broader variety of scenes, even when fewer edges can be observed. Our experiments show that our algorithms achieve comparable results to the state of the art in general and produce superior outputs on man-made scenes and photos degraded by a small kernel.

In blind deconvolution one aims to estimate from an input blurred image y a sharp image x and an unknown blur kernel k. Recent research shows that a key to success is to consider the overall shape of the posterior distribution p(x, k\y) and not only its mode. This leads to a distinction between MAPx, kstrategies which estimate the mode pair x, k and often lead to undesired results, and MAPkstrategies which select the best k while marginalizing over all possible x images. The MAPkprinciple is significantly more robust than the MAPx, kone, yet, it involves a challenging marginalization over latent images. As a result, MAPktechniques are considered complicated, and have not been widely exploited. This paper derives a simple approximated MAPkalgorithm which involves only a modest modification of common MAPx, kalgorithms. We show that MAPkcan, in fact, be optimized easily, with no additional computational complexity.

Motions can occur over both short and long time scales. We introduce motion denoising, which treats short-term changes as noise, long-term changes as signal, and re-renders a video to reveal the underlying long-term events. We demonstrate motion denoising for time-lapse videos. One of the characteristics of traditional time-lapse imagery is stylized jerkiness, where short-term changes in the scene appear as small and annoying jitters in the video, often obfuscating the underlying temporal events of interest. We apply motion denoising for resynthesizing time-lapse videos showing the long-term evolution of a scene with jerky short-term changes removed. We show that existing filtering approaches are often incapable of achieving this task, and present a novel computational approach to denoise motion without explicit motion analysis. We demonstrate promising experimental results on a set of challenging time-lapse sequences.

Diffuse objects generally tell us little about the surrounding lighting, since the radiance they reflect blurs together incident lighting from many directions. In this paper we discuss how occlusion geometry can help invert diffuse reflectance to recover lighting or surface albedo. Self-occlusion in the scene can be regarded as a form of coding, creating high frequencies that improve the conditioning of diffuse light transport. Our analysis builds on a basic observation that diffuse reflectors with sufficiently detailed geometry can fully resolve the incident lighting. Using a Bayesian framework, we propose a novel reconstruction method based on high-resolution photography, taking advantage of visibility changes near occlusion boundaries. We also explore the limits of single-pixel observations as the diffuse reflector (and potentially the lighting) vary over time. Diffuse reflectance imaging is particularly relevant for astronomy applications, where diffuse reflectors arise naturally but the incident lighting and camera position cannot be controlled. To test our approaches, we first study the feasibility of using the moon as a diffuse reflector to observe the earth as seen from space. Next we present a reconstruction of Mars using historical photometry measurements not previously used for this purpose. As our results suggest, diffuse reflectance imaging expands our notion of what can qualify as a camera.

Image features are widely used in computer vision applications. They need to be robust to scene changes and image transformations. Designing and comparing feature descriptors requires the ability to evaluate their performance with respect to those transformations. We want to know how robust the descriptors are to changes in the lighting, scene, or viewing conditions. For this, we need ground truth data of different scenes viewed under different camera or lighting conditions in a controlled way. Such data is very difficult to gather in a real-world setting. We propose using a photorealistic virtual world to gain complete and repeatable control of the environment in order to evaluate image features. We calibrate our virtual world evaluations by comparing against feature rankings made from photographic data of the same subject matter (the Statue of Liberty). We find very similar feature rankings between the two datasets. We then use our virtual world to study the effects on descriptor performance of controlled changes in viewpoint and illumination. We also study the effect of augmenting the descriptors with depth information to improve performance.

This paper describes areas and problems where computer vision can use help from the discrete algorithms community.

Blind deconvolution is the recovery of a sharp version of a blurred image when the blur kernel is unknown. Recent algorithms have afforded dramatic progress, yet many aspects of the problem remain challenging and hard to understand. The goal of this paper is to analyze and evaluate recent blind deconvolution algorithms both theoretically and experimentally. We explain the previously reported failure of the naive MAP approach by demonstrating that it mostly favors no-blur explanations. We show that, using reasonable image priors, a naive simulations MAP estimation of both latent image and blur kernel is guaranteed to fail even with infinitely large images sampled from the prior. On the other hand, we show that since the kernel size is often smaller than the image size, a MAP estimation of the kernel alone is well constrained and is guaranteed to succeed to recover the true blur. The plethora of recent deconvolution techniques makes an experimental evaluation on ground-truth data important. As a first step toward this experimental evaluation, we have collected blur data with ground truth and compared recent algorithms under equal settings. Additionally, our data demonstrate that the shift-invariant blur assumption made by most algorithms is often violated.

Computer-generated (CG) images have achieved high levels of realism. This realism, however, comes at the cost of long and expensive manual modeling, and often humans can still distinguish between CG and real images. We introduce a new data-driven approach for rendering realistic imagery that uses a large collection of photographs gathered from online repositories. Given a CG image, we retrieve a small number of real images with similar global structure. We identify corresponding regions between the CG and real images using a mean-shift cosegmentation algorithm. The user can then automatically transfer color, tone, and texture from matching regions to the CG image. Our system only uses image processing operations and does not require a 3D model of the scene, making it fast and easy to integrate into digital content creation workflows. Results of a user study show that our hybrid images appear more realistic than the originals.

Blur is caused by a pixel receiving light from multiple scene points, and in many cases, such as object motion, the induced blur varies spatially across the image plane. However, the seemingly straight-forward task of estimating spatially-varying blur from a single image has proved hard to accomplish reliably. This work considers such blur and makes two contributions: a local blur cue that measures the likelihood of a small neighborhood being blurred by a candidate blur kernel; and an algorithm that, given an image, simultaneously selects a motion blur kernel and segments the region that it affects. The methods are shown to perform well on a diversity of images.

We explore the problem of reconstructing an image from a bag of square, non-overlapping image patches, the jigsaw puzzle problem. Completing jigsaw puzzles is challenging and requires expertise even for humans, and is known to be NP-complete. We depart from previous methods that treat the problem as a constraint satisfaction problem and develop a graphical model to solve it. Each patch location is a node and each patch is a label at nodes in the graph. A graphical model requires a pairwise compatibility term, which measures an affinity between two neighboring patches, and a local evidence term, which we lack. This paper discusses ways to obtain these terms for the jigsaw puzzle problem. We evaluate several patch compatibility metrics, including the natural image statistics measure, and experimentally show that the dissimilarity-based compatibility - measuring the sum-of-squared color difference along the abutting boundary - gives the best results. We compare two forms of local evidence for the graphical model: a sparse-and-accurate evidence and a dense-and-noisy evidence. We show that the sparse-and-accurate evidence, fixing as few as 4 - 6 patches at their correct locations, is enough to reconstruct images consisting of over 400 patches. To the best of our knowledge, this is the largest puzzle solved in the literature. We also show that one can coarsely estimate the low resolution image from a bag of patches, suggesting that a bag of image patches encodes some geometric information about the original image.

In image restoration tasks, a heavy-tailed gradient distribution of natural images has been extensively exploited as an image prior. Most image restoration algorithms impose a sparse gradient prior on the whole image, reconstructing an image with piecewise smooth characteristics. While the sparse gradient prior removes ringing and noise artifacts, it also tends to remove mid-frequency textures, degrading the visual quality. We can attribute such degradations to imposing an incorrect image prior. The gradient profile in fractal-like textures, such as trees, is close to a Gaussian distribution, and small gradients from such regions are severely penalized by the sparse gradient prior. To address this issue, we introduce an image restoration algorithm that adapts the image prior to the underlying texture. We adapt the prior to both low-level local structures as well as mid-level textural characteristics. Improvements in visual quality is demonstrated on deconvolution and denoising tasks.

Taking multiple exposures is a well-established approach both for capturing high dynamic range (HDR) scenes and for noise reduction. But what is the optimal set of photos to capture? The typical approach to HDR capture uses a set of photos with geometrically-spaced exposure times, at a fixed ISO setting (typically ISO 100 or 200). By contrast, we show that the capture sequence with optimal worst-case performance, in general, uses much higher and variable ISO settings, and spends longer capturing the dark parts of the scene. Based on a detailed model of noise, we show that optimal capture can be formulated as a mixed integer programming problem. Compared to typical HDR capture, our method lets us achieve higher worst-case SNR in the same capture time (for some cameras, up to 19 dB improvement in the darkest regions), or much faster capture for the same minimum acceptable level of SNR. Our experiments demonstrate this advantage for both real and synthetic scenes.

We propose Recursive Compositional Models (RCMs) for simultaneous multi-view multi-object detection and parsing (e.g. view estimation and determining the positions of the object subparts). We represent the set of objects by a family of RCMs where each RCM is a probability distribution defined over a hierarchical graph which corresponds to a specific object and viewpoint. An RCM is constructed from a hierarchy of subparts/subgraphs which are learnt from training data. Part-sharing is used so that different RCMs are encouraged to share subparts/subgraphs which yields a compact representation for the set of objects and which enables efficient inference and learning from a limited number of training samples. In addition, we use appearance-sharing so that RCMs for the same object, but different viewpoints, share similar appearance cues which also helps efficient learning. RCMs lead to a multi-view multi-object detection system. We illustrate RCMs on four public datasets and achieve state-of-the-art performance.

We present a latent hierarchical structural learning method for object detection. An object is represented by a mixture of hierarchical tree models where the nodes represent object parts. The nodes can move spatially to allow both local and global shape deformations. The models can be trained discriminatively using latent structural SVM learning, where the latent variables are the node positions and the mixture component. But current learning methods are slow, due to the large number of parameters and latent variables, and have been restricted to hierarchies with two layers. In this paper we describe an incremental concave-convex procedure (iCCCP) which allows us to learn both two and three layer models efficiently. We show that iCCCP leads to a simple training algorithm which avoids complex multi-stage layer-wise training, careful part selection, and achieves good performance without requiring elaborate initialization. We perform object detection using our learnt models and obtain performance comparable with state-of-the-art methods when evaluated on challenging public PASCAL datasets. We demonstrate the advantages of three layer hierarchies - outperforming Felzenszwalb et al.'s two layer models on all 20 classes.

The patch transform represents an image as a bag of overlapping patches sampled on a regular grid. This representation allows users to manipulate images in the patch domain, which then seeds the inverse patch transform to synthesize modified images. Possible modifications include the spatial locations of patches, the size of the output image, or the pool of patches from which an image is reconstructed. When no modifications are made, the inverse patch transform reduces to solving a jigsaw puzzle. The inverse patch transform is posed as a patch assignment problem on a Markov random field (MRF), where each patch should be used only once and neighboring patches should fit to form a plausible image. We find an approximate solution to the MRF using loopy belief propagation, introducing an approximation that encourages the solution to use each patch only once. The image reconstruction algorithm scales well with the total number of patches through label pruning. In addition, structural misalignment artifacts are suppressed through a patch jittering scheme that spatially jitters the assigned patches. We demonstrate the patch transform and its effectiveness on natural images.

We present a system for generating “infinite” images from large collections of photos by means of transformed image retrieval. Given a query image, we first transform it to simulate how it would look if the camera moved sideways and then perform image retrieval based on the transformed image. We then blend the query and retrieved images to create a larger panorama. Repeating this process will produce an “infinite” image. The transformed image retrieval model is not limited to simple 2-D left/right image translation, however, and we show how to approximate other camera motions like rotation and forward motion/zoom-in using simple 2-D image transforms. We represent images in the database as a graph where each node is an image and different types of edges correspond to different types of geometric transformations simulating different camera motions. Generating infinite images is thus reduced to following paths in the image graph. Given this data structure we can also generate a panorama that connects two query images, simply by finding the shortest path between the two in the image graph. We call this option the “image taxi.” Our approach does not assume photographs are of a single real 3-D location, nor that they were taken at the same time. Instead, we organize the photos in themes, such as city streets or skylines and synthesize new virtual scenes by combining images from distinct but visually similar locations. There are a number of potential applications to this technology. It can be used to generate long panoramas as well as content aware transitions between reference images or video shots. Finally, the image graph allows users to interactively explore large photo collections for ideation, games, social interaction, and artistic purposes.

Blind deconvolution is the recovery of a sharp version of a blurred image when the blur kernel is unknown. Recent algorithms have afforded dramatic progress, yet many aspects of the problem remain challenging and hard to understand. The goal of this paper is to analyze and evaluate recent blind deconvolution algorithms both theoretically and experimentally. We explain the previously reported failure of the naive MAP approach by demonstrating that it mostly favors no-blur explanations. On the other hand we show that since the kernel size is often smaller than the image size a MAP estimation of the kernel alone can be well constrained and accurately recover the true blur. The plethora of recent deconvolution techniques makes an experimental evaluation on ground-truth data important. We have collected blur data with ground truth and compared recent algorithms under equal settings. Additionally, our data demonstrates that the shift-invariant blur assumption made by most algorithms is often violated.

The intrinsic image decomposition aims to retrieve “intrinsic” properties of an image, such as shading and reflectance. To make it possible to quantitatively compare different approaches to this problem in realistic settings, we present a ground-truth dataset of intrinsic image decompositions for a variety of real-world objects. For each object, we separate an image of it into three components: Lambertian shading, reflectance, and specularities. We use our dataset to quantitatively compare several existing algorithms; we hope that this dataset will serve as a means for evaluating future work on intrinsic images.

Capturing multiple photos at different focus settings is a powerful approach for reducing optical blur, but how many photos should we capture within a fixed time budget? We develop a framework to analyze optimal capture strategies balancing the tradeoff between defocus and sensor noise, incorporating uncertainty in resolving scene depth. We derive analytic formulas for restoration error and use Monte Carlo integration over depth to derive optimal capture strategies for different camera designs, under a wide range of photographic scenarios. We also derive a new upper bound on how well spatial frequencies can be preserved over the depth of field. Our results show that by capturing the optimal number of photos, a standard camera can achieve performance at the level of more complex computational cameras, in all but the most demanding of cases. We also show that computational cameras, although specifically designed to improve one-shot performance, generally benefit from capturing multiple photos as well.

The theory of compressed sensing tells a dramatic story that sparse signals can be reconstructed near-perfectly from a small number of random measurements. However, recent work has found the story to be more complicated. For example, the projections based on principal component analysis work better than random projections for some images while the reverse is true for other images. Which feature of images makes such a distinction and what is the optimal set of projections for natural images? In this paper, we attempt to answer these questions with a novel formulation of compressed sensing. In particular, we find that bandwise random projections in which more projections are allocated to low spatial frequencies are near-optimal for natural images and demonstrate using experimental results that the bandwise random projections outperform other kinds of projections in image reconstruction.

In this paper, we investigate how similar images sharing the same global description can help with unsupervised scene segmentation in an image. In contrast to recent work in semantic alignment of scenes, we allow an input image to be explained by partial matches of similar scenes. This allows for a better explanation of the input scenes. We perform MRF-based segmentation that optimizes over matches, while respecting boundary information. The recovered segments are then used to re-query a large database of images to retrieve better matches for the target region. We show improved performance in detecting occluding boundaries over previous methods on data gathered from the LabelMe database.

We present a nonparametric Bayesian method for texture learning and synthesis. A texture image is represented by a 2D-Hidden Markov Model (2D-HMM) where the hidden states correspond to the cluster labeling of textons and the transition matrix encodes their spatial layout (the compatibility between adjacent textons). 2D-HMM is coupled with the Hierarchical Dirichlet process (HDP) which allows the number of textons and the complexity of transition matrix grow as the input texture becomes irregular. The HDP makes use of Dirichlet process prior which favors regular textures by penalizing the model complexity. This framework (HDP-2D-HMM) learns the texton vocabulary and their spatial layout jointly and automatically. The HDP-2D-HMM results in a compact representation of textures which allows fast texture synthesis with comparable rendering quality over the state-of-the-art image-based rendering methods. We also show that HDP-2D-HMM can be applied to perform image segmentation and synthesis.

Depth of field (DOF), the range of scene depths that appear sharp in a photograph, poses a fundamental tradeoff in photography---wide apertures are important to reduce imaging noise, but they also increase defocus blur. Recent advances in computational imaging modify the acquisition process to extend the DOF through deconvolution. Because deconvolution quality is a tight function of the frequency power spectrum of the defocus kernel, designs with high spectra are desirable. In this paper we study how to design effective extended-DOF systems, and show an upper bound on the maximal power spectrum that can be achieved. We analyze defocus kernels in the 4D light field space and show that in the frequency domain, only a low-dimensional 3D manifold contributes to focus. Thus, to maximize the defocus spectrum, imaging systems should concentrate their limited energy on this manifold. We review several computational imaging systems and show either that they spend energy outside the focal manifold or do not achieve a high spectrum over the DOF. Guided by this analysis we introduce the lattice-focal lens, which concentrates energy at the low-dimensional focal manifold and achieves a higher power spectrum than previous designs. We have built a prototype lattice-focal lens and present extended depth of field results.

We introduce the patch transform, where an image is broken into non-overlapping patches, and modifications or constraints are applied in the “patch domain”. A modified image is then reconstructed from the patches, subject to those constraints. When no constraints are given, the reconstruction problem reduces to solving a jigsaw puzzle. Constraints the user may specify include the spatial locations of patches, the size of the output image, or the pool of patches from which an image is reconstructed. We define terms in a Markov network to specify a good image reconstruction from patches: neighboring patches must fit to form a plausible image, and each patch should be used only once. We find an approximate solution to the Markov network using loopy belief propagation, introducing an approximation to handle the combinatorially difficult patch exclusion constraint. The resulting image reconstructions show the original image, modified to respect the user’s changes. We apply the patch transform to various image editing tasks and show that the algorithm performs well on real world images.

Obtaining ground-truth motion for arbitrary, real-world video sequences is a challenging but important task for both algorithm evaluation and model design. Existing ground-truth databases are either synthetic, such as the Yosemite sequence, or limited to indoor, experimental setups, such as the database developed by Baker et al (2007). We propose a human-in-loop methodology to create a ground-truth motion database for the videos taken with ordinary cameras in both indoor and outdoor scenes, using the fact that human beings are experts at segmenting objects and inspecting the match between two frames. We designed an interactive computer vision system to allow a user to efficiently annotate motion. Our methodology is cross-validated by showing that human annotated motion is repeatable, consistent across annotators, and close to the ground truth obtained by Baker et al (2007). Using our system, we collected and annotated 10 indoor and outdoor real-world videos to form a ground-truth motion database. The source code, annotation tool and database is online for public evaluation and benchmarking.

Objects in the world can be arranged into a hierarchy based on their semantic meaning (e.g. organism - animal - feline - cat). What about defining a hierarchy based on the visual appearance of objects? This paper investigates ways to automatically discover a hierarchical structure for the visual world from a collection of unlabeled images. Previous approaches for unsupervised object and scene discovery focused on partitioning the visual data into a set of non-overlapping classes of equal granularity. In this work, we propose to group visual objects using a multi-layer hierarchy tree that is based on common visual elements. This is achieved by adapting to the visual domain the generative hierarchical latent Dirichlet allocation (hLDA) model previously used for unsupervised discovery of topic hierarchies in text. Images are modeled using quantized local image regions as analogues to words in text. Employing the multiple segmentation framework of Russell et al. [22], we show that meaningful object hierarchies, together with object segmentations, can be automatically learned from unlabeled and unsegmented image collections without supervision. We demonstrate improved object classification and localization performance using hLDA over the previous non-hierarchical method on the MSRC dataset [33].

Computational Vision and Machine Learning have become synergistic fields of research.Modern machine learning techniques have improved the state of the art in computer vision and catalized re-thinking of key problems, such as recognition and tracking.In turn, vision has broadened the scope of machine learning, offering rich new challenges and highlighting the importance of representations.This special issue contains 15 papers at the intersection of vision and learning, most firmly rooted in both areas.The topics span a range of computer vision topics, from object recognition and tracking to lower-level visual tasks such as feature detection and finding contours and segments.The machine learning methods include discriminative and generative models, as well as both parametric and non-parametric representations.Since databases are proving to be crucial for progress in both machine learning and computer vision, we

Image denoising algorithms often assume an additive white Gaussian noise (AWGN) process that is independent of the actual RGB values. Such approaches are not fully automatic and cannot effectively remove color noise produced by todays CCD digital camera. In this paper, we propose a unified framework for two tasks: automatic estimation and removal of color noise from a single image using piecewise smooth image models. We introduce the noise level function (NLF), which is a continuous function describing the noise level as a function of image brightness. We then estimate an upper bound of the real noise level function by fitting a lower envelope to the standard deviations of per-segment image variances. For denoising, the chrominance of color noise is significantly removed by projecting pixel values onto a line fit to the RGB values in each segment. Then, a Gaussian conditional random field (GCRF) is constructed to obtain the underlying clean image from the noisy input. Extensive experiments are conducted to test the proposed algorithm, which is shown to outperform state-of-the-art denoising algorithms.

With the advent of the Internet, billions of images are now freely available online and constitute a dense sampling of the visual world. Using a variety of non-parametric methods, we explore this world with the aid of a large dataset of 79,302,017 images collected from the Internet. Motivated by psychophysical results showing the remarkable tolerance of the human visual system to degradations in image resolution, the images in the dataset are stored as 32 x 32 color images. Each image is loosely labeled with one of the 75,062 non-abstract nouns in English, as listed in the Wordnet lexical database. Hence the image database gives a comprehensive coverage of all object categories and scenes. The semantic information from Wordnet can be used in conjunction with nearest-neighbor methods to perform object classification over a range of semantic levels minimizing the effects of labeling noise. For certain classes that are particularly prevalent in the dataset, such as people, we are able to demonstrate a recognition performance comparable to class-specific Viola-Jones style detectors.

Object motion during camera exposure often leads to noticeable blurring artifacts. Proper elimination of this blur is challenging because the blur kernel is unknown, varies over the image as a function of object velocity, and destroys high frequencies. In the case of motions along a 1D direction (e.g. horizontal) we show that these challenges can be addressed using a camera that moves during the exposure. Through the analysis of motion blur as space-time integration, we show that a parabolic integration (corresponding to constant sensor acceleration) leads to motion blur that is invariant to object velocity. Thus, a single deconvolution kernel can be used to remove blur and create sharp images of scenes with objects moving at different speeds, without requiring any segmentation and without knowledge of the object speeds. Apart from motion invariance, we prove that the derived parabolic motion preserves image frequency content nearly optimally. That is, while static objects are degraded relative to their image from a static camera, a reliable reconstruction of all moving objects within a given velocities range is made possible. We have built a prototype camera and present successful deblurring results over a wide variety of human motions.

Markov random field (MRF) models are a popular tool for vision and image processing. Gaussian MRF models are particularly convenient to work with because they can be implemented using matrix and linear algebra routines. However, recent research has focused on on discrete-valued and non-convex MRF models because Gaussian models tend to over-smooth images and blur edges. In this paper, we show how to train a Gaussian conditional random field (GCRF) model that overcomes this weakness and can outperform the non-convex field of experts model on the task of denoising images. A key advantage of the GCRF model is that the parameters of the model can be optimized efficiently on relatively large images. The competitive performance of the GCRF model and the ease of optimizing its parameters make the GCRF model an attractive option for vision and image processing applications.

Many low-level vision algorithms assume a prior probability over images, and there has been great interest in trying to learn this prior from examples. Since images are very non Gaussian, high dimensional, continuous signals, learning their distribution presents a tremendous computational challenge. Perhaps the most successful recent algorithm is the Fields of Experts (FOE) [20] model which has shown impressive performance by modeling image statistics with a product of potentials defined on filter outputs. However, as in previous models of images based on filter outputs [30], calculating the probability of an image given the model requires evaluating an intractable partition function. This makes learning very slow (requires Monte-Carlo sampling at every step) and makes it virtually impossible to compare the likelihood of two different models. Given this computational difficulty, it is hard to say whether nonintu-itive features learned by such models represent a true property of natural images or an artifact of the approximations used during learning. In this paper we present (1) tractable lower and upper bounds on the partition function of models based on filter outputs and (2) efficient learning algorithms that do not require any sampling. Our results are based on recent results in machine learning that deal with Gaussian potentials. We extend these results to non-Gaussian potentials and derive a novel, basis rotation algorithm for approximating the maximum likelihood filters. Our results allow us to (1) rigorously compare the likelihood of different models and (2) calculate high likelihood models of natural image statistics in a matter of minutes. Applying our results to previous models shows that the nonintuitive features are not an artifact of the learning process but rather are capturing robust properties of natural images.

Mole pattern changes are important cues in detecting melanoma at an early stage. As a first step to automatically register mole pattern changes from skin images, this paper presents a framework to detect and label moles on skin images in the presence of clutter, occlusions, and varying imaging conditions. The input image is processed with cascaded blocks to successively discard non-mole pixels. Our method first searches the entire input image for skin regions using a non-parametric skin detection scheme, and the detected skin regions are further processed using a difference of Gaussian (DoG) filter to find possible mole candidates of varying sizes. Mole candidates are classified as moles in the final stage using a trained support vector machine. To increase the mole classification accuracy, hair is removed if present on the skin image using steerable filters and a graphical model. The performance of the designed system is evaluated with 28 test images, and the experimental results demonstrate the effectiveness of the proposed mole localization scheme.

Current object recognition systems can only recognize a limited number of object categories; scaling up to many categories is the next challenge. We seek to build a system to recognize and localize many different object categories in complex scenes. We achieve this through a simple approach: by matching the input im- age, in an appropriate representation, to images in a large training set of labeled images. Due to regularities in object identities across similar scenes, the retrieved matches provide hypotheses for object identities and locations. We build a prob- abilistic model to transfer the labels from the retrieval set to the input image. We demonstrate the effectiveness of this approach and study algorithm component contributions using held-out test sets from the LabelMe database.

We consider the problem of detecting a large number of different classes of objects in cluttered scenes. Traditional approaches require applying a battery of different classifiers to the image, at multiple locations and scales. This can be slow and can require a lot of training data since each classifier requires the computation of many different image features. In particular, for independently trained detectors, the (runtime) computational complexity and the (training-time) sample complexity scale linearly with the number of classes to be detected. We present a multitask learning procedure, based on boosted decision stumps, that reduces the computational and sample complexity by finding common features that can be shared across the classes (and/or views). The detectors for each class are trained jointly, rather than independently. For a given performance level, the total number of features required and, therefore, the runtime cost of the classifier, is observed to scale approximately logarithmically with the number of classes. The features selected by joint training are generic edge-like features, whereas the features chosen by training each class separately tend to be more object-specific. The generic features generalize better and considerably reduce the computational cost of multiclass object detection.

A conventional camera captures blurred versions of scene information away from the plane of focus. Camera systems have been proposed that allow for recording all-focus images, or for extracting depth, but to record both simultaneously has required more extensive hardware and reduced spatial resolution. We propose a simple modification to a conventional camera that allows for the simultaneous recovery of both (a) high resolution image information and (b) depth information adequate for semi-automatic extraction of a layered depth representation of the image. Our modification is to insert a patterned occluder within the aperture of the camera lens, creating a coded aperture. We introduce a criterion for depth discriminability which we use to design the preferred aperture pattern. Using a statistical model of images, we can recover both depth information and an all-focus image from single photographs taken with the modified camera. A layered depth map is then extracted, requiring user-drawn strokes to clarify layer assignments in some cases. The resulting sharp image and layered depth map can be combined for various photographic applications, including automatic scene segmentation, post-exposure refocusing, or re-rendering of the scene from an alternate viewpoint.

In order to work well, many computer vision algorithms require that their parameters be adjusted according to the image noise level, making it an important quantity to estimate. We show how to estimate an upper bound on the noise level from a single image based on a piecewise smooth image prior model and measured CCD camera response functions. We also learn the space of noise level functions how noise level changes with respect to brightness and use Bayesian MAP inference to infer the noise level function from a single image. We illustrate the utility of this noise estimation for two algorithms: edge detection and featurepreserving smoothing through bilateral filtering. For a variety of different noise levels, we obtain good results for both these algorithms with no user-specified inputs.

Given a large dataset of images, we seek to automatically determine the visually similar object and scene classes together with their image segmentation. To achieve this we combine two ideas: (i) that a set of segmented objects can be partitioned into visual object classes using topic discovery models from statistical text analysis; and (ii) that visual object classes can be used to assess the accuracy of a segmentation. To tie these ideas together we compute multiple segmentations of each image and then: (i) learn the object classes; and (ii) choose the correct segmentations. We demonstrate that such an algorithm succeeds in automatically discovering many familiar objects in a variety of image datasets, including those from Caltech, MSRC and LabelMe.

We develop an integrated, probabilistic model for the appearance and three-dimensional geometry of cluttered scenes. Object categories are modeled via distributions over the 3D location and appearance of visual features. Uncertainty in the number of object instances depicted in a particular image is then achieved via a transformed Dirichlet process. In contrast with image-based approaches to object recognition, we model scale variations as the perspective projection of objects in different 3D poses. To calibrate the underlying geometry, we incorporate binocular stereo images into the training process. A robust likelihood model accounts for outliers in matched stereo features, allowing effective learning of 3D object structure from partial 2D segmentations. Applied to a dataset of office scenes, our model detects objects at multiple scales via a coarse reconstruction of the corresponding 3D geometry.

Images can be represented as the composition of multiple intrinsic component images, such as shading, albedo, and noise images. In this paper, we present a method for estimating intrinsic component images from a single image, which we apply to the problems of estimating shading and albedo images and image denoising. Our method is based on learning estimators that predict filtered versions of the desired image. Unlike previous approaches, our method does not require unnatural discretizations of the problem. We also demonstrate how to learn a weighting function that properly weights the local estimates when constructing the estimated image. For shading estimation, we introduce a new training set of real-world images. The accuracy of our method is measured both qualitatively and quantitatively, showing better performance on the shading/albedo separation problem than previous approaches. The performance on denoising is competitive with the current state of the art.

A reliable motion estimation algorithm must function under a wide range of con- ditions. One regime, which we consider here, is the case of moving objects with contours but no visible texture. Tracking distinctive features such as corners can disambiguate the motion of contours, but spurious features such as T-junctions can be badly misleading. It is difficult to determine the reliability of motion from local measurements, since a full rank covariance matrix can result from both real and spurious features. We propose a novel approach that avoids these points al- together, and derives global motion estimates by utilizing information from three levels of contour analysis: edgelets, boundary fragments and contours. Boundary fragment are chains of orientated edgelets, for which we derive motion estimates from local evidence. The uncertainties of the local estimates are disambiguated after the boundary fragments are properly grouped into contours. The grouping is done by constructing a graphical model and marginalizing it using importance sampling. We propose two equivalent representations in this graphical model, re- versible switch variables attached to the ends of fragments and fragment chains, to capture both local and global statistics of boundaries. Our system is success- fully applied to both synthetic and real video sequences containing high-contrast boundaries and textureless regions. The system produces good motion estimates along with properly grouped and completed contours.

Camera shake during exposure leads to objectionable image blur and ruins many photographs. Conventional blind deconvolution methods typically assume frequency-domain constraints on images, or overly simplified parametric forms for the motion path during camera shake. Real camera motions can follow convoluted paths, and a spatial domain prior can better maintain visually salient image characteristics. We introduce a method to remove the effects of camera shake from seriously blurred images. The method assumes a uniform camera blur over the image and negligible in-plane camera rotation. In order to estimate the blur from the camera shake, the user must specify an image region without saturation effects. We show results for a variety of digital photographs taken from personal photo collections.

In cross-modal inference, we estimate complete fields from noisy and missing observations of one sensory modality using structure found in another sensory modality. This inference problem occurs in several areas including texture reconstruction and reconstruction of geophysical fields. We propose a method for cross-modal inference that simultaneously learns shape recipes between two modalities and estimates missing information by using a prior on image structure gleaned from the alternate modality. In the absence of a physical basis for representing image priors, we use a statistical one that represents correlations in differential features. This is done efficiently using a perturbation sampling scheme. Using just one example of the alternate modality, we produce a factorized ensemble representation of feature correlations that yields efficient solutions to large-sized spatial inference problems. We demonstrate the utility of this approach on cross-modal inference with depth and spectral data.

We seek to discover the object categories depicted in a set of unlabelled images. We achieve this using a model developed in the statistical text literature: probabilistic latent semantic analysis (pLSA). In text analysis, this is used to discover topics in a corpus using the bag-of-words document representation. Here we treat object categories as topics, so that an image containing instances of several categories is modeled as a mixture of topics. The model is applied to images by using a visual analogue of a word, formed by vector quantizing SIFT-like region descriptors. The topic discovery approach successfully translates to the visual domain: for a small set of objects, we show that both the object categories and their approximate spatial layout are found without supervision. Performance of this unsupervised method is compared to the supervised approach of Fergus et al. (2003) on a set of unseen images containing only one object per image. We also extend the bag-of-words vocabulary to include 'doublets' which encode spatially local co-occurring regions. It is demonstrated that this extended vocabulary gives a cleaner image segmentation. Finally, the classification and segmentation methods are applied to a set of images containing multiple objects per image. These results demonstrate that we can successfully build object class models from an unsupervised analysis of images.

We describe a hierarchical probabilistic model for the detection and recognition of objects in cluttered, natural scenes. The model is based on a set of parts which describe the expected appearance and position, in an object centered coordinate frame, of features detected by a low-level interest operator. Each object category then has its own distribution over these parts, which are shared between objects. We learn the parameters of this model via a Gibbs sampler which uses the graphical model's structure to analytically average over many parameters. Applied to a database of images of isolated objects, the sharing of parts among objects improves detection accuracy when few training examples are available. We also extend this hierarchical framework to scenes containing multiple objects

Motivated by the problem of learning to detect and recognize objects with minimal supervision, we develop a hierarchical probabilistic model for the spatial structure of visual scenes. In contrast with most existing models, our approach explicitly captures uncertainty in the number of object instances depicted in a given image. Our scene model is based on the transformed Dirichlet process (TDP), a novel extension of the hierarchical DP in which a set of stochastically transformed mixture components are shared between multiple groups of data. For visual scenes, mixture components describe the spatial structure of visual features in an objectcentered coordinate frame, while transformations model the object positions in a particular image. Learning and inference in the TDP, which has many potential applications beyond computer vision, is based on an empirically effective Gibbs sampler. Applied to a dataset of partially labeled street scenes, we show that the TDP's inclusion of spatial structure improves detection performance, flexibly exploiting partially labeled training images.

Interpreting real-world images requires the ability distinguish the different characteristics of the scene that lead to its final appearance. Two of the most important of these characteristics are the shading and reflectance of each point in the scene. We present an algorithm that uses multiple cues to recover shading and reflectance intrinsic images from a single image. Using both color information and a classifier trained to recognize gray-scale patterns, given the lighting direction, each image derivative is classified as being caused by shading or a change in the surface's reflectance. The classifiers gather local evidence about the surface's form and color, which is then propagated using the generalized belief propagation algorithm. The propagation step disambiguates areas of the image where the correct classification is not clear from local evidence. We use real-world images to demonstrate results and show how each component of the system affects the results.

Important inference problems in statistical physics, computer vision, error-correcting coding theory, and artificial intelligence can all be reformulated as the computation of marginal probabilities on factor graphs. The belief propagation (BP) algorithm is an efficient way to solve these problems that is exact when the factor graph is a tree, but only approximate when the factor graph has cycles. We show that BP fixed points correspond to the stationary points of the Bethe approximation of the free energy for a factor graph. We explain how to obtain region-based free energy approximations that improve the Bethe approximation, and corresponding generalized belief propagation (GBP) algorithms. We emphasize the conditions a free energy approximation must satisfy in order to be a "valid" or "maxent-normal" approximation. We describe the relationship between four different methods that can be used to generate valid approximations: the "Bethe method", the "junction graph method", the "cluster variation method", and the "region graph method". Finally, we explain how to tell whether a region-based approximation, and its corresponding GBP algorithm, is likely to be accurate, and describe empirical results showing that GBP can significantly outperform BP.

We present motion magnification, a technique that acts like a microscope for visual motion. It can amplify subtle motions in a video sequence, allowing for visualization of deformations that would otherwise be invisible. To achieve motion magnification, we need to accurately measure visual motions, and group the pixels to be modified. After an initial image registration step, we measure motion by a robust analysis of feature point trajectories, and segment pixels based on similarity of position, color, and motion. A novel measure of motion similarity groups even very small motions according to correlation over time, which often relates to physical cause. An outlier mask marks observations not explained by our layered motion model, and those pixels are simply reproduced on the output from the original registered observations.The motion of any selected layer may be magnified by a user-specified amount; texture synthesis fills-in unseen "holes" revealed by the amplified motions. The resulting motion-magnified images can reveal or emphasize small motions in the original sequence, as we demonstrate with deformations in load-bearing structures, subtle motions or balancing corrections of people, and "rigid" structures bending under hand pressure.

We address the problem of text super-resolution: given a single image of text scanned in at low resolution from a piece of paper, return the image that is mostly likely to be generated from a noiseless high-resolution scan of the same piece of paper. In doing so, we wish to: (1) avoid introducing artifacts in the high-resolution image such as blurry edges and rounded corners, (2) recover from quantization noise and grid-alignment effects that introduce errors in the low-resolution image, and (3) handle documents with very large glyph sets such as Japanese's Kanji. Applications for this technology include improving the display of: fax documents, low-resolution scans of archival documents, and low-resolution bitmapped fonts on high-resolution output devices.

We describe a threedimensional geometric hand model suitable for vi- sual tracking applications. The kinematic constraints implied by the model's joints have a probabilistic structure which is well described by a graphical model. Inference in this model is complicated by the hand's many degrees of freedom, as well as multimodal likelihoods caused by ambiguous image measurements. We use nonparametric belief propaga- tion (NBP) to develop a tracking algorithm which exploits the graph's structure to control complexity, while avoiding costly discretization. While kinematic constraints naturally have a local structure, self occlusions created by the imaging process lead to complex interpenden- cies in color and edgebased likelihood functions. However, we show that local structure may be recovered by introducing binary hidden vari- ables describing the occlusion state of each pixel. We augment the NBP algorithm to infer these occlusion variables in a distributed fashion, and then analytically marginalize over them to produce hand position esti- mates which properly account for occlusion events. We provide simula- tions showing that NBP may be used to refine inaccurate model initializa- tions, as well as track hand motion through extended image sequences. 1 Introduction Accurate visual detection and tracking of threedimensional articulated objects is a chal- lenging problem with applications in humancomputer interfaces, motion capture, and scene understanding [1]. In this paper, we develop a probabilistic method for tracking a geometric hand model from monocular image sequences. Because articulated hand mod- els have many (roughly 26) degrees of freedom, exact representation of the posterior dis- tribution over model configurations is intractable. Trackers based on extended and un- scented Kalman filters [2, 3] have difficulties with the multimodal uncertainties produced by ambiguous image evidence. This has motived many researchers to consider nonparamet- ric representations, including particle filters [4, 5] and deterministic multiscale discretiza- tions [6]. However, the hand's high dimensionality can cause these trackers to suffer catas- trophic failures, requiring the use of models which limit the hand's motion [4] or sophisti- cated prior models of hand configurations and dynamics [5, 6]. An alternative way to address the high dimensionality of articulated tracking problems is to describe the posterior distribution's statistical structure using a graphical model. Graph- Figure 1: Projected edges (left block) and silhouettes (right block) for a configuration of the 3D structural hand model matching the given image. To aid visualization, the model is also projected following rotations by 35 (center) and 70 (right) about the vertical axis. ical models have been used to track viewbased human body representations [7], con- tour models of restricted hand configurations [8], viewbased 2.5D "cardboard" models of hands and people [9], and a full 3D kinematic human body model [10]. Because the variables in these graphical models are continuous, and discretization is intractable for threedimensional models, most traditional graphical inference algorithms are inapplica- ble. Instead, these trackers are based on recently proposed extensions of particle filters to general graphs: mean field Monte Carlo in [9], and nonparametric belief propagation (NBP) [11, 12] in [10]. In this paper, we show that NBP may be used to track a threedimensional geometric model of the hand. To derive a graphical model for the tracking problem, we consider a redun- dant local representation in which each hand component is described by its own three dimensional position and orientation. We show that the model's kinematic constraints, including selfintersection constraints not captured by joint angle representations, take a simple form in this local representation. We also provide a local decomposition of the likelihood function which properly handles occlusion in a distributed fashion, a significant improvement over our earlier tracking results [13]. We conclude with simulations demon- strating our algorithm's robustness to occlusions. 2 Geometric Hand Modeling Structurally, the hand is composed of sixteen approximately rigid components: three pha- langes or links for each finger and thumb, as well as the palm [1]. As proposed by [2, 3], we model each rigid body by one or more truncated quadrics (ellipsoids, cones, and cylin- ders) of fixed size. These geometric primitives are well matched to the true geometry of the hand, allow tracking from arbitrary orientations (in contrast to 2.5D "cardboard" mod- els [5, 9]), and permit efficient computation of projected boundaries and silhouettes [3]. Figure 1 shows the edges and silhouettes corresponding to a sample hand model configu- ration. Note that only a coarse model of the hand's geometry is necessary for tracking. 2.1 Kinematic Representation and Constraints The kinematic constraints between different hand model components are well described by revolute joints [1]. Figure 2(a) shows a graph describing this kinematic structure, in which nodes correspond to rigid bodies and edges to joints. The two joints connecting the phalanges of each finger and thumb have a single rotational degree of freedom, while the joints connecting the base of each finger to the palm have two degrees of freedom (cor- responding to grasping and spreading motions). These twenty angles, combined with the palm's global position and orientation, provide 26 degrees of freedom. Forward kinematic transformations may be used to determine the finger positions corresponding to a given set of joint angles. While most modelbased hand trackers use this joint angle parameteriza- tion, we instead explore a redundant representation in which the ith rigid body is described by its position qi and orientation ri (a unit quaternion). Let xi = (qi, ri) denote this local description of each component, and x = {x1, . . . , x16} the overall hand configuration. Clearly, there are dependencies among the elements of x implied by the kinematic con- (a) (b) (c) (d) Figure 2: Graphs describing the hand model's constraints. (a) Kinematic constraints (EK ) de- rived from revolute joints. (b) Structural constraints (ES) preventing 3D component intersections. (c) Dynamics relating two consecutive time steps. (d) Occlusion consistency constraints (EO). straints. Let EK be the set of all pairs of rigid bodies which are connected by joints, or equivalently the edges in the kinematic graph of Fig. 2(a). For each joint (i, j) EK , define an indicator function K (x i,j i, xj ) which is equal to one if the pair (xi, xj ) are valid rigid body configurations associated with some setting of the angles of joint (i, j), and zero otherwise. Viewing the component configurations xi as random variables, the following prior explicitly enforces all constraints implied by the original joint angle representation: pK(x) K (x i,j i, xj ) (1) (i,j)EK Equation (1) shows that pK (x) is an undirected graphical model, whose Markov structure is described by the graph representing the hand's kinematic structure (Fig. 2(a)). 2.2 Structural and Temporal Constraints In reality, the hand's joint angles are coupled because different fingers can never occupy the same physical volume. This constraint is complex in a joint angle parameterization, but simple in our local representation: the position and orientation of every pair of rigid bodies must be such that their component quadric surfaces do not intersect. We approximate this ideal constraint in two ways. First, we only explicitly constrain those pairs of rigid bodies which are most likely to intersect, corresponding to the edges ES of the graph in Fig. 2(b). Furthermore, because the relative orientations of each finger's quadrics are implicitly constrained by the kinematic prior pK (x), we may detect most intersections based on the distance between object centroids. The structural prior is then given by 1 ||q p i - qj || > i,j S (x) S (x (x i,j i, xj ) S i,j i, xj ) = (2) 0 otherwise (i,j)ES where i,j is determined from the quadrics composing rigid bodies i and j. Empirically, we find that this constraint helps prevent different fingers from tracking the same image data. In order to track hand motion, we must model the hand's dynamics. Let xt denote the i position and orientation of the ith hand component at time t, and xt = {xt1, . . . , xt16}. For each component at time t, our dynamical model adds a Gaussian potential connecting it to the corresponding component at the previous time step (see Fig. 2(c)): 16 pT xt | xt-1 = N xt - xt-1; 0, i i i (3) i=1 Although this temporal model is factorized, the kinematic constraints at the following time step implicitly couple the corresponding random walks. These dynamics can be justified as the maximum entropy model given observations of the nodes' marginal variances i. 3 Observation Model Skin colored pixels have predictable statistics, which we model using a histogram distribu- tion pskin estimated from training patches [14]. Images without people were used to create a histogram model pbkgd of nonskin pixels. Let (x) denote the silhouette of projected hand configuration x. Then, assuming pixels are independent, an image y has likelihood p p skin(u) C (y | x) = pskin(u) pbkgd(v) (4) pbkgd(u) u(x) v(x) u(x) The final expression neglects the proportionality constant p v bkgd(v), which is inde- pendent of x, and thereby limits computation to the silhouette region [8]. 3.1 Distributed Occlusion Reasoning In configurations where there is no selfocclusion, pC (y | x) decomposes as a product of local likelihood terms involving the projections (xi) of individual hand components [13]. To allow a similar decomposition (and hence distributed inference) when there is occlu- sion, we augment the configuration xi of each node with a set of binary hidden variables zi = {zi } = 0 if pixel u in the projection of rigid body i is occluded (u) u. Letting zi(u) by any other body, and 1 otherwise, the color likelihood (eq. (4)) may be rewritten as 16 p z 16 i(u) p skin(u) C (y | x, z) = = p p C (y | xi, zi) (5) bkgd(u) i=1 u(xi) i=1 Assuming they are set consistently with the hand configuration x, the hidden occlusion variables z ensure that the likelihood of each pixel in (x) is counted exactly once. We may enforce consistency of the occlusion variables using the following function: 0 if x = 1 (x j occludes xi, u (xj ), and zi(u) j , zi ; x (6) (u) i) = 1 otherwise Note that because our rigid bodies are convex and nonintersecting, they can never take mutually occluding configurations. The constraint (xj, zi ; x (u) i) is zero precisely when pixel u in the projection of xi should be occluded by xj, but zi is in the unoccluded state. (u) The following potential encodes all of the occlusion relationships between nodes i and j: O (x (x ; x ; x i,j i, zi, xj , zj ) = j , zi(u) i) (xi, zj(u) j ) (7) u These occlusion constraints exist between all pairs of nodes. As with the structural prior, we enforce only those pairs EO (see Fig. 2(d)) most prone to xj occlusion: pO(x, z) O (x i,j i, zi, xj , zj ) (8) (i,j)EO z y i(u) xi Figure 3 shows a factor graph for the occlusion relationships between xi and its neighbors, as well as the observation potential pC (y | xi, zi). x u k The occlusion potential (xj, zi ; x (u) i) has a very Figure 3: Factor graph showing weak dependence on xi, depending only on p(y | xi, zi), and the occlusion con- whether xi is behind xj relative to the camera. straints placed on xi by xj , xk. Dashed lines denote weak dependencies. The 3.2 Modeling Edge Filter Responses plate is replicated once per pixel. Edges provide another important hand tracking cue. Using boundaries labeled in training images, we estimated a histogram pon of the response of a derivative of Gaussian filter steered to the edge's orientation [8, 10]. A similar histogram poff was estimated for filter outputs at randomly chosen locations. Let (x) denote the oriented edges in the projection of model configuration x. Then, again assuming pixel independence, image y has edge likelihood p 16 p z 16 i(u) p on(u) on(u) E (y | x, z) = = p p E (y | xi, zi) (9) off (u) poff(u) u(x) i=1 u(xi) i=1 where we have used the same occlusion variables z to allow a local decomposition. 4 Nonparametric Belief Propagation Over the previous sections, we have shown that a redundant, local representation of the geometric hand model's configuration xt allows p (xt | yt), the posterior distribution of the hand model at time t given image observations yt, to be written as 16 p xt | yt pK(xt)pS(xt)pO(xt, zt) pC(yt | xt, zt)p , zt) i i E (yt | xti i (10) zt i=1 The summation marginalizes over the hidden occlusion variables zt, which were needed to locally decompose the edge and color likelihoods. When video frames are observed, the overall posterior distribution is given by p (x | y) p xt | yt pT (xt | xt-1) (11) t=1 Excluding the potentials involving occlusion variables, which we discuss in detail in Sec. 4.2, eq. (11) is an example of a pairwise Markov random field: p (x | y) i,j (xi, xj) i (xi, y) (12) (i,j)E iV Hand tracking can thus be posed as inference in a graphical model, a problem we propose to solve using belief propagation (BP) [15]. At each BP iteration, some node i V calculates a message m (x ij j ) to be sent to a neighbor j (i) {j | (i, j) E}: mn (x mn-1 (x ij j ) j,i (xj , xi) i (xi, y) ki i) dxi (13) xi k(i)\j At any iteration, each node can produce an approximation ^ p(xi | y) to the marginal distri- bution p (xi | y) by combining the incoming messages with the local observation: ^ pn(xi | y) i (xi, yi) mn (x ji i) (14) j(i) For treestructured graphs, the beliefs ^ pn(xi | y) will converge to the true marginals p (xi | y). On graphs with cycles, BP is approximate but often highly accurate [15]. 4.1 Nonparametric Representations For the hand tracking problem, the rigid body configurations xi are sixdimensional con- tinuous variables, making accurate discretization intractable. Instead, we employ nonpara- metric, particlebased approximations to these messages using the nonparametric belief propagation (NBP) algorithm [11, 12]. In NBP, each message is represented using either a samplebased density estimate (a mixture of Gaussians) or an analytic function. Both types of messages are needed for hand tracking, as we discuss below. Each NBP message update involves two stages: sampling from the estimated marginal, followed by Monte Carlo ap- proximation of the outgoing message. For the general form of these updates, see [11]; the following sections focus on the details of the hand tracking implementation. The hand tracking application is complicated by the fact that the orientation component ri of xi = (qi, ri) is an element of the rotation group SO(3). Following [10], we represent orientations as unit quaternions, and use a linearized approximation when constructing den- sity estimates, projecting samples back to the unit sphere as necessary. This approximation is most appropriate for densities with tightly concentrated rotational components. 4.2 Marginal Computation BP's estimate of the belief ^ p(xi | y) is equal to the product of the incoming messages from neighboring nodes with the local observation potential (see eq. (14)). NBP approximates this product using importance sampling, as detailed in [13] for cases where there is no selfocclusion. First, M samples are drawn from the product of the incoming kinematic and temporal messages, which are Gaussian mixtures. We use a recently proposed multi- scale Gibbs sampler [16] to efficiently draw accurate (albeit approximate) samples, while avoiding the exponential cost associated with direct sampling (a product of d M Gaussian mixtures contains M d Gaussians). Following normalization of the rotational component, each sample is assigned a weight equal to the product of the color and edge likelihoods with any structural messages. Finally, the computationally efficient "rule of thumb" heuris- tic [17] is used to set the bandwidth of Gaussian kernels placed around each sample. To derive BP updates for the occlusion masks zi, we first cluster (xi, zi) for each hand component so that p (xt, zt | yt) has a pairwise form (as in eq. (12)). In principle, NBP could manage occlusion constraints by sampling candidate occlusion masks zi along with rigid body configurations xi. However, due to the exponentially large number of possible occlusion masks, we employ a more efficient analytic approximation. Consider the BP message sent from xj to (zi, xi), calculated by applying eq. (13) to the occlusion potential (x ; x u j , zi(u) i). We assume that ^ p(xj | y) is well separated from any candidate xi, a situation typically ensured by the kinematic and structural constraints. The occlusion constraint's weak dependence on xi (see Fig. 3) then separates the message computation into two cases. If xi lies in front of typical xj configurations, the BP message j,i(u)(zi ) is uninformative. If x (u) i is occluded, the message approximately equals j,i(u)(zi = 0) = 1 = 1) = 1 - Pr [u (x (u) j,i(u)(zi(u) j )] (15) where we have neglected correlations among pixel occlusion states, and where the prob- ability is computed with respect to ^ p(xj | y). By taking the product of these messages k,i(u)(zi ) from all potential occluders x (u) k and normalizing, we may determine an ap- proximation to the marginal occlusion probability i Pr[z = 0]. (u) i(u) Because the color likelihood pC (y | xi, zi) factorizes across pixels u, the BP approximation to pC (y | xi) may be written in terms of these marginal occlusion probabilites: p p skin(u) C (y | xi) i + (1 - ) (16) (u) i(u) pbkgd(u) u(xi) Intuitively, this equation downweights the color evidence at pixel u as the probability of that pixel's occlusion increases. The edge likelihood pE(y | xi) averages over zi similarly. The NBP estimate of ^ p(xi | y) is determined by sampling configurations of xi as before, and reweighting them using these occlusionsensitive likelihood functions. 4.3 Message Propagation To derive the propagation rule for nonocclusion edges, as suggested by [18] we rewrite the message update equation (13) in terms of the marginal distribution ^ p(xi | y): ^ pn-1(x mn (x i | y) dx ij j ) = j,i (xj , xi) i (17) x mn-1 (x i ji i) Our explicit use of the current marginal estimate ^ pn-1(xi | y) helps focus the Monte Carlo approximation on the most important regions of the state space. Note that messages sent 1 2 1 2 Figure 4: Refinement of a coarse initialization following one and two NBP iterations, both without (left) and with (right) occlusion reasoning. Each plot shows the projection of the five most significant modes of the estimated marginal distributions. Note the difference in middle finger estimates. along kinematic, structural, and temporal edges depend only on the belief ^ p(xi | y) follow- ing marginalization over occlusion variables zi. Details and pseudocode for the message propagation step are provided in [13]. For kine- matic constraints, we sample uniformly among permissable joint angles, and then use forward kinematics to propagate samples from ^ pn-1(xi | y) /mn-1 (x ji i) to hypothesized configurations of xj. Following [12], temporal messages are determined by adjusting the bandwidths of the current marginal estimate ^ p(xi | y) to match the temporal covariance i. Because structural potentials (eq. (2)) equal one for all state configurations outside some ball, the ideal structural messages are not finitely integrable. We therefore approximate the structural message m (x ij j ) as an analytic function equal to the weights of all kernels in ^ p(xi | y) outside a ball centered at qj, the position of xj.

We seek to both detect and segment objects in images. To exploit both lo- cal image data as well as contextual information, we introduce Boosted Random Fields (BRFs), which uses Boosting to learn the graph struc- ture and local evidence of a conditional random field (CRF). The graph structure is learned by assembling graph fragments in an additive model. The connections between individual pixels are not very informative, but by using dense graphs, we can pool information from large regions of the image; dense models also support efficient inference. We show how contextual information from other objects can improve detection perfor- mance, both in terms of accuracy and speed, by using a computational cascade. We apply our system to detect stuff and things in office and street scenes. 1 Introduction Our long-term goal is to build a vision system that can examine an image and describe what objects are in it, and where. In many images, such as Fig. 5(a), objects of interest, such as the keyboard or mouse, are so small that they are impossible to detect just by using local features. Seeing a blob next to a keyboard, humans can infer it is likely to be a mouse; we want to give a computer the same abilities. There are several pieces of related work. Murphy et al [9] used global scene context to help object recognition, but did not model relationships between objects. Fink and Perona [4] exploited local dependencies in a boosting framework, but did not allow for multiple rounds of communication between correlated objects. He et al [6] do not model connections between objects directly, but rather they induce such correlations indirectly, via a bank of hidden variables, using a "restricted Boltzmann machine" architecture. In this paper, we exploit contextual correlations between the object classes by introducing Boosted Random Fields (BRFs). Boosted random fields build on both boosting [5, 10] and conditional random fields (CRFs) [8, 7, 6]. Boosting is a simple way of sequentially constructing "strong" classifiers from "weak" components, and has been used for single- class object detection with great success [12]. Dietterich et al [3] combine boosting and 1D CRFs, but they only consider the problem of learning the local evidence potentials; we consider the much harder problem of learning the structure of a 2D CRF. Standard applications of MRFs/ CRFs to images [7] assume a 4-nearest neighbor grid structure. While successful in low-level vision, this structure will fail in capturing im- portant long distance dependencies between whole regions and across classes. We propose a method for learning densely connected random fields with long range connections. The topology of these connections is chosen by a weak learner which has access to a library of graph fragments, derived from patches of labeled training images, which reflect typical spatial arrangments of objects (similar to the segmentation fragments in [2]). At each round of the learning algorithm, we add more connections from other locations in the image and from other classes (detectors). The connections are assumed to be spatially invariant, which means this update can be performed using convolution followed by a sigmoid nonlinearity. The resulting architecture is similar to a convolutional neural network, although we used a stagewise training procedure, which is much faster than back propagation. In addition to recognizing things, such as cars and people, we are also interested in recog- nizing spatially extended "stuff" [1], such as roads and buildings. The traditional sliding window approach to object detection does not work well for detecting "stuff". Instead, we combine object detection and image segmentation (c.f., [2]) by labeling every pixel in the image. We do not rely on a bottom-up image segmentation algorithm, which can be fragile without top-down guidance. 2 Learning potentials and graph structure A conditional random field (CRF) is a distribution of the form 1 P (S|x) = Z i(Si) i,j (Si, Sj ) i jNi where x is the input (e.g., image), Ni are the neighbors of node i, and Si are labels. We have assumed pairwise potentials for notational simplicity. Our goal is to learn the local evidence potentials, i, the compatibility potentials , and the set of neighbors Ni. We propose the following simple approximation: use belief propagation (BP) to estimate the marginals, P (Si|x), and then use boosting to maximize the likelihood of each node's training data with respect to i and . In more detail, the algorithm is as follows. At iteration t, the goal is to minimize the negative log-likelihood of the training data. As in [11], we consider the per-label loss (i.e., we use marginal probabilities), as opposed to requiring that the joint labeling be correct (as in Viterbi decoding). Hence the cost function to be minimized is Jt = Jti = - bti,m(Si,m) = - bti,m(+1)Si,mbti,m(-1)1-Si,m (1) i m i m i where Si,m {-1, +1} is the true label for pixel i in training case m, Si,m = (Si,m + 1)/2 {0, 1} is just a relabeling, and bti,m = [P (Si = -1|xm, t), P (Si = 1|xm, t)] is the belief state at node i given input image xm after t iterations of the algorithm. The belief at node i is given by the following (dropping the dependence on case m) bti(1) ti(1) Mti(1) where Mti is the product of all the messages coming into i from all its neighbors at time t and where the message that k sends to i is given by bt (s M t+1(1) = t+1 (1) t+1 (1) = k k) i (2) ki ki k,i(sk, 1) t (sk) kN ik i sk{-1,+1} where k,i is the compatility between nodes k and i. If we assume that the local potentials have the form t /2 /2 i(si) = [eF t i ; e-F ti ], where F ti is some function of the input data, then: bti(+1) = (F ti + Gti), Gti = log Mti(+1) - log Mti(-1) (3) where (u) = 1/(1 + e-u) is the sigmoid function. Hence each term in Eq. 1 simplifies to a cost function similar to that used in boosting: log Jt +Gt ) i,m i = log 1 + e-Si,m(F ti,m . (4) m 1. Input: a set of labeled pairs {xi,m; Si,m}, bound T Output: Local evidence functions f ti(x) and message update functions gti(bN ). i 2. Initialize: bt=0 i,m = 0; F t=0 i,m = 0; Gt=0 i,m = 0 3. For t=1..T. (a) Fit local potential fi(xi,m) by weighted LS to Y t +Gt ) i,m i,m = Si,m(1 + e-Si,m(F t i ) (b) .Fit compatibilities gti(bt-1 ) to Y t N i,m by weighted LS. i ,m (c) Compute local potential F t i,m = F t-1 + f t i,m i (xi,m) (d) Compute compatibilities Gti,m = t gn ) n=1 i (bt-1 Ni,m (e) Update the beliefs bti,m = (F ti,m + Gti,m) (f) Update weights wt+1 = bt i,m i,m(-1) bt i,m(+1) Figure 1: BRF training algorithm. We assume that the graph is very densely connected so that the information that one single node sends to another is so small that we can make the approximation t+1 (+1)/ t+1 (-1) 1. (This is a reasonable approximation in the case of images, ki ki where each node represents a single pixel; only when the influence of many pixels is taken into account will the messages become informative.) Hence bt (s k,m k ) M t+1(+1) s k,i(sk, +1) t (s Gt+1 = log i = log k [-1,+1] i k ) k i (5) M t+1(-1) bt (sk) i k,m k s k,i(sk, -1) k [-1,+1] t (s i k ) k k,i(sk, +1) bt (s k,m k) log sk[-1,+1] (6) k,i(sk, -1) bt (sk) k sk[-1,+1] k,m With this simplification, Gt+1 (bt i is now a non-linear function of the beliefs Gt+1 i m) at iteration t. Therefore, We can write the beliefs at iteration t as a function of the local evidences and the beliefs at time t - 1: bti(+1) = (F ti(xi,m) + Gti(bt-1 m )). The key idea behind BRFs is to use boosting to learn the G functions, which approximately implement message passing in densely connected graphs. We explain this in more detail below. 2.1 Learning local evidence potentials Defining F ti(xi,m) = F t-1(x i i,m) + f t i (xi,m) as an additive model, where xi,m are the features of training sample m at node i, we can learn this function in a stagewise fashion by optimizing the second order Taylor expansion of Eq. 4 wrt f ti, as in logitBoost [5]: arg min log Jti arg min wti,m(Y ti,m - fti(xi,m))2 (7) f t f t i i m where Y t +Gt ) i,m i,m = Si,m(1+e-Si,m(F t i ). In the case that the weak learner is a "regression stump", fi(x) = ah(x)+b, we can find the optimal a, b by solving a weighted least squares problem, with weights wti,m = bti(-1) bti(+1); we can find the best basis function h(x) by searching over all elements of a dictionary. 2.2 Learning compatibility potentials and graph structure In this section, we discuss how to learn the compatibility functions ij, and hence the structure of the graph. Instead of learning the compatibility functions ij, we propose to 1. Input: a set of inputs {xi,m} and functions f ti, gti Output: Set of beliefs bi,m and MAP estimates Si,m. 2. Initialize: bt=0 i,m = 0; F t=0 i,m = 0; Gt=0 i,m = 0 3. From t = 1 to T , repeat (a) Update local evidences F t i,m = F t-1 + f t i,m i (xi,m) (b) Update compatibilities Gti,m = t gn ) n=1 i (bt-1 Ni,m (c) Compute current beliefs bti,m = (F ti,m + Gti,m) 4. Output classification is Si,m = bti,m > 0.5 Figure 2: BRF run-time inference algorithm. learn directly the function Gt+1 i . We propose to use an additive model for Gt+1 i as we did for learning F : Gt+1 = t gn i,m n=1 i (btm), where btm is a vector with the beliefs of all nodes in the graph at iteration t for the training sample m. The weak learners gn i (btm) can be regression stumps with the form gn i (btm) = a(w btm > ) + b, where a, b, are the parameters of the regression stump, and wi is a set of weights selected from a dictionary. In the case of a graph with weak and almost symmetrical connections (which holds if (s1, s2) 1, for all (s1, s2), which implies the messages are not very informative) we can further simplify the function Gt+1 i by approximating it as a linear function of the beliefs: Gt+1 = i,m k,i btk,m(+1) + k,i (8) kNi This step reduces the computational cost. The weak learners gn i (btm) will also be linear functions. Hence the belief update simplifies to bt+1(+1) = ( i,m i btm + i + F t i,m), which is similar to the mean-field update equations. The neighborhood Ni over which we sum incoming messages is determined by the graph structure, which is encoded in the non-zero values of i. Each weak learner gn i will compute a weighted combination of the beliefs of the some subset of the nodes; this subset may change from iteration to iteration, and can be quite large. At iteration t, we choose the weak learner gti so as to minimize t-1 log Jt +gt(bt-1)+ gn(bt-1)) i m i m i (bt-1) = - log 1 + e-Si,m(F ti,m n=1 m which reduces to a weighted least squares problem similar to Eq. 7. See Fig. 1 for the pseudo-code for the complete learning algorithm, and Fig. 2 for the pseudo-code for run- time inference. 3 BRFs for multiclass object detection and segmentation With the BRF training algorithm in hand, we describe our approach for multiclass object detection and region-labeling using densely connected BRFs. 3.1 Weak learners for detecting stuff and things The square sliding window approach does not provide a natural way of working with irreg- ular objects. Using region labeling as an image representation allows dealing with irregular and extended objects (buildings, bookshelf, road, ...). Extended stuff [1] may be a very important source of contextual information for other objects. (a) Examples from the dictionary of about 2000 patches and masks, Ux,y, Vx,y. (b) Examples from the dictionary of 30 graphs, Wx,y,c. f t=0 f t=1 f t=2 F S + + ... = put thu utO Tr (c) Example feedforward segmentation for screens. Figure 3: Examples of patches from the dictionary and an example of the segmentation obtained using boosting trained with patches from (a). The weak learners we use for the local evidence potentials are based on the segmentation fragments proposed in [2]. Specifically, we create a dictionary of about 2000 image patches U , chosen at random (but overlapping each object), plus a corresponding set of binary (in- class/ out-of-class) image masks, V : see Fig. 3(a). At each round t, for each class c, and for each dictionary entry, we construct the following weak learner, whose output is a binary matrix of the same size as the image I: v(I) = ((I U ) > ) V > 0 (9) where represents normalized cross-correlation and represents convolution. The in- tuition behind this is that I U will produce peaks at image locations that contain this patch/template, and then convolving with V will superimpose the segmentation mask on top of the peaks. As a function of the threshold , the feature will behave more as a template detector ( 1) or as a texture descriptor ( << 1). To be able to detect objects at multiple scales, we first downsample the image to scale , compute v(I ), and then upsample the result. The final weak learner does this for multiple scales, ORs all the results together, and then takes a linear transformation. f (I) = ([v(I ) ]) + (10) Fig. 3(c) shows an example of segmentation obtained by using boosting without context. The weak learners we use for the compatibility functions have a similar form: C gc(b) = bc Wc + (11) c=1 where bc is the image formed by the beliefs at all pixels for class c. This convolution corresponds to eq. 8 in which the node i is one pixel x, y of class c. The binary kernels (graph fragments) W define, for each node x, y of object class c, all the nodes from which it will receive messages. These kernels are chosen by sampling patches of various sizes from the labeling of images from the training set. This allows generating complicated patterns of connectivity that reflect the statistics of object co-occurrences in the training set. The overall incoming message is given by adding the kernels obtained at each boosting round. (This is the key difference from mutual boosting [4], where the incoming message is just the output of a single weak learner; thus, in mutual boosting, previously learned inter-class connections are only used once.) Although it would seem to take O(t) time to compute Gt, we can precompute a single equivalent kernel W , so at runtime the overall complexity is still linear in the number of boosting rounds, O(T ). C t C Gtx,y,c = bc nW n c + ndef = b + c W c c=1 n=1 n c=1 car car building car road car Road F b=(F+G) Car car building building building road building Building x G car road building road road road y c) A car out of context a) Incoming messages (outside 3rd floor windows) to a car node. b) Compatibilities (W'). is less of a car. t=1 t=2 t=4 t=20 t=40 Final labeling b(car) S(all) d) Evolution of the beliefs for the car nodes (b) and labeling (S) for road, building, car. Figure 4: Street scene. The BRF is trained to detect cars, buildings and the road. In Fig. 4(a-b), we show the structures of the graph and the weights W defined by GT for a BRF trained to detect cars, buildings and roads in street scenes. 3.2 Learning and inference For training we used a labeled dataset of office and street scenes with about 100 images in each set. During the training, in the first 5 rounds we only update the local potentials, to allow local evidence to accrue. After the 5th iteration we start updating also the compatibil- ity functions. At each round, we update only the local potential and compatibility function associated with a single object class that reduces the most the multiclass cost. This allows objects that need many features to have more complicated local potentials. The algorithm learns to first detect easy (and large) objects, since these reduce the error of all classes the fastest. The easy-to-detect objects can then pass information to the harder ones. For instance, in office scenes, the system first detects screens, then keyboards, and finally computer mice. Fig. 5 illustrates this behavior on the test set. A similar behavior is obtained for the car detector (Fig. 4(d)). The detection of building and road provides strong constraints for the locations of the car. 3.3 Cascade of classifiers with BRFs The BRF can be turned into a cascade [12] by thresholding the beliefs. Computations can then be reduced by doing the convolutions (required for computing f and g) only in pixels that are still candidates for the presence of the target. At each round we update a binary rejection mask for each object class, Rtx,y,c, by thresholding the beliefs at round t: Rtx,y,c = Rt-1 x,y,c (btx,y,c > tc). A pixel in the rejection mask is set to zero when we can decide that the object is not present (when btx,y,c is below the threshold tc 0), and it is set to 1 when more processing is required. The threshold tc is chosen so that the percentage of missed detections is below a predefined level (we use 1%). Similarity we can define a detection mask that will indicate pixels in which we decide the object is present. The mask is then used for computing the features v(I) and messages G by applying the convolutions only on the pixels not yet classified. We can denote those operators as R and R. This Input image screen mouse Ground truth Output labeling keyboard t=5 t=10 t=15 t=25 t=50 b (screen) b (screen) b (screen) b (screen) b (screen) F G b (keyboard) b (keyboard) b (keyboard) b (keyboard) b (keyboard) F G b (mouse) b (mouse) b (mouse) b (mouse) b (mouse) F G 1 ROC Screen Boosting BRF Mouse a under Keyboard re Iteration (t) A 0.5 t=0 t=20 t=50 Figure 5: Top. In this desk scene, it is easy to identify objects like the screen, keyboard and mouse, even though the local information is sometimes insufficient. Middle: the evolution of the beliefs (b and F and G) during detection for a test image. Bottom. The graph bellow shows the average evolution of the area under the ROC for the three objects on 120 test images. results in a more efficient classifier with only a slight decrease of performance. In Fig. 6 we compare the reduction of the search space when implementing a cascade using independent boosting (which reduces to Viola and Jones [12]), and when using BRF's. We see that for objects for which context is the main source of information, like the mouse, the reduction in search space is much more dramatic using BRFs than using boosting alone. 4 Conclusion The proposed BRF algorithm combines boosting and CRF's, providing an algorithm that is easy for both training and inference. We have demonstrated object detection in cluttered scenes by exploiting contextual relationships between objects. The BRF algorithm is com- putationally efficient and provides a natural extension of the cascade of classifiers by inte- grating evidence from other objects in order to quickly reject certain image regions. The BRF's densely connected graphs, which efficiently collect information over large image regions, provide an alternative framework to nearest-neighbor grids for vision problems.

We introduce a new method to describe shape relationships over time in a photograph. We acquire both range and image information in a sequence of frames using a stationary stereo camera. From the pictures taken, we compute a composite image consisting of the pixels from the surfaces closest to the camera over all the time frames. Through occlusion cues, this composite reveals 3-D relationships between the shapes at different times. We call the composite a shape-time photograph. Small errors in stereo depth measurements can create artifacts in the shape-time images. We correct most of these using a Markov network to estimate the most probable front-surface pixel, taking into account (a) the stereo depth measurements and their uncertainties, and (b) spatial continuity assumptions for the time-frame assignments of the front-surface pixels.

In many applications of graphical models arising in computer vision, the hidden variables of interest are most naturally specified by continuous, non-Gaussian distributions. There exist inference algorithms for discrete approximations to these continuous distributions, but for the high-dimensional variables typically of interest, discrete inference becomes infeasible. Stochastic methods such as particle filters provide an appealing alternative. However, existing techniques fail to exploit the rich structure of the graphical models describing many vision problems. Drawing on ideas from regularized particle filters and belief propagation (BP), this paper develops a nonparametric belief propagation (NBP) algorithm applicable to general graphs. Each NBP iteration uses an efficient sampling procedure to update kernel-based approximations to the true, continuous likelihoods. The algorithm can accommodate an extremely broad class of potential functions, including nonparametric representations. Thus, NBP extends particle filtering methods to the more general vision problems that graphical models can describe. We apply the NBP algorithm to infer component interrelationships in a parts-based face model, allowing location and reconstruction of occluded features.

In low-level vision, the representation of scene properties such as shape, albedo, etc., are very high dimensional as they have to describe complicated structures. The approach proposed here is to let the image itself bear as much of the representational burden as possible. In many situations, scene and image are closely related and it is possible to find a functional relationship between them. The scene information can be represented in reference to the image where the functional specifies how to translate the image into the associated scene. We illustrate the use of this representation for encoding shape information. We show how this representation has appealing properties such as locality and slow variation across space and scale. These properties provide a way of improving shape estimates coming from other sources of information like stereo.

Recent stereo algorithms have achieved impressive results by modelling the disparity image as a Markov Random Field (MRF). An important component of an MRF-based approach is the inference algorithm used to find the most likely setting of each node in the MRF. Algorithms have been proposed which use graph cuts or belief propagation for inference. These stereo algorithms differ in both the inference algorithm used and the formulation of the MRF. It is unknown whether to attribute the responsibility for differences in performance to the MRF or the inference algorithm. We address this through controlled experiments by comparing the belief propagation algorithm and the graph cuts algorithm on the same MRF's, which have been created for calculating stereo disparities. We find that the labellings produced by the two algorithms are comparable. The solutions produced by graph cuts have a lower energy than those produced with belief propagation, but this does not necessarily lead to increased performance relative to the ground truth.

While navigating in an environment, a vision system has to be able to recognize where it is and what the main objects in the scene are. We present a context-based vision system for place and object recognition. The goal is to identify familiar locations (e.g., office 610, conference room 941, main street), to categorize new environments (office, corridor, street) and to use that information to provide contextual priors for object recognition (e.g., tables are more likely in an office than a street). We present a low-dimensional global image representation that provides relevant information for place recognition and categorization, and show how such contextual information introduces strong priors that simplify object recognition. We have trained the system to recognize over 60 locations (indoors and outdoors) and to suggest the presence and locations of more than 20 different object types. The algorithm has been integrated into a mobile system that provides realtime feedback to the user.

The problem of approximating the product of several Gaussian mixture distributions arises in a number of contexts, including the nonparametric belief propagation (NBP) inference algorithm and the training of prod- uct of experts models. This paper develops two multiscale algorithms for sampling from a product of Gaussian mixtures, and compares their performance to existing methods. The first is a multiscale variant of pre- viously proposed Monte Carlo techniques, with comparable theoretical guarantees but improved empirical convergence rates. The second makes use of approximate kernel density evaluation methods to construct a fast approximate sampler, which is guaranteed to sample points to within a tunable parameter (cid:15) of their true probability. We compare both multi- scale samplers on a set of computational examples motivated by NBP, demonstrating significant improvements over existing methods.

Standard approaches to object detection focus on local patches of the image, and try to classify them as background or not. We propose to use the scene context (image as a whole) as an extra source of (global) information, to help resolve local ambiguities. We present a conditional random field for jointly solving the tasks of object detection and scene classification.

THE last 10 years have witnessed rapid growth in the popularity of graphical models, most notably Bayesian networks, as a tool for representing, learning, and computing complex probability distributions. Graphical models provide an explicit representation of the statistical dependencies between the components of a complex probability model, effectively marrying probability theory and graph theory. As Jordan puts it in [2], graphical models are “a natural tool for dealing with two problems that occur throughout applied mathematics and engineering—uncertainty and complexity—and, in particular, they are playing an increasingly important role in the design and analysis of machine learning algorithms.” Graphical models provide powerful computational support for the Bayesian approach to computer vision, which has become a standard framework for addressing vision problems. Many familiar tools from the vision literature, such as Markov random fields, hidden Markov models, and the Kalman filter, are instances of graphical models. More importantly, the graphical models formalism makes it possible to generalize these tools and develop novel statistical representations and associated algorithms for inference and learning. The history of graphical models in computer vision follows closely that of graphical models in general. Research by Pearl [3] and Lauritzen [4] in the late 1980s played a seminal role in introducing this formalism to areas of AI and statistical learning. Not long after, the formalism spread to fields such as statistics, systems engineering, information theory, pattern recognition, and, among others, computer vision. One of the earliest occurrences of graphical models in the vision literature was a paper by Binford et al. [1]. The paper described the use of Bayesian inference in a hierarchical probability model to match 3D object models to groupings of curves in a single image. The following year marked the publication of Pearl’s influential book [3] on graphical models. Since then, many technical papers have been published in IEEE journals and conference proceedings that address different aspects and applications of graphical models in computer vision. Our goal in organizing this special section was to demonstrate the breadth of applicability of the graphical models formalism to vision problems. Our call for papers in February 2002 produced 16 submissions. After a careful review process, we selected six papers for publication, including five regular papers, and one short paper. These papers reflect the state-of-the-art in the use of graphical models in vision problems that range from low-level image understanding to high-level scene interpretation. We believe these papers will appeal both to vision researchers who are actively engaged in the use of graphical models and machine learning researchers looking for a challenging application domain. The first paper in this section is “Stereo Matching Using Belief Propagation” by J. Sun, N.-N. Zheng, and H.-Y. Shum. The authors describe a new stereo algorithm based on loopy belief propagation, a powerful inference technique for complex graphical models in which exact inference is intractable. They formulate the dense stereo matching problem as MAP estimation on coupled Markov random fields and obtain promising results on standard test data sets. One of the benefits of this formulation, as the authors demonstrate, is the ease with which it can be extended to handle multiview stereo matching. In their paper “Statistical Cue Integration of DAG Deformable Models” S.K. Goldenstein, C. Vogler, and D. Metaxas describe a scheme for combining different sources of information into estimates of the parameters of a deformable model. They use a DAG representation of the interdependencies between the nodes in a deformable model. This framework supports the efficient integration of information from edges and other cues using the machinery of affine arithmetic and the propagation of uncertainties. They present experimental results for a face tracking application. Y. Song, L. Goncalves, and P. Perona describe, in their paper “Unsupervised Learning of Human Motion,” a method for learning probabilistic models of human motion from video sequences in cluttered scenes. Two key advantages of their method are its unsupervised nature, which can mitigate the need for tedious hand labeling of data, and the utilization of graphical model constraints to reduce the search space when fitting a human figure model. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 25, NO. 7, JULY 2003 785

We present an example-based method for translating line drawings into different styles. We fit each line as a linear combination of similar lines in a training set, and interpolate between the corresponding training examples in the output style. The synthesized lines preserve the desired stylistic features of the output style.

The goal of low-level vision is to estimate an underlying scene, given an observed image. Real-world scenes (eg, albedos or shapes) can be very complex, conventionally requiring high dimensional representations which are hard to estimate and store. We propose a low-dimensional rep- resentation, called a scene recipe, that relies on the image itself to de- scribe the complex scene configurations. Shape recipes are an example: these are the regression coefficients that predict the bandpassed shape from image data. We describe the benefits of this representation, and show two uses illustrating their properties: (1) we improve stereo shape estimates by learning shape recipes at low resolution and applying them at full resolution; (2) Shape recipes implicitly contain information about lighting and materials and we use them for material segmentation.

We present an algorithm that uses multiple cues to recover shading and reflectance intrinsic images from a single image. Using both color in- formation and a classifier trained to recognize gray-scale patterns, each image derivative is classified as being caused by shading or a change in the surface’s reflectance. Generalized Belief Propagation is then used to propagate information from areas where the correct classification is clear to areas where it is ambiguous. We also show results on real images.

We introduce general-education students to key ideas in applied computing through case studies from computer graphics, computer animation, image processing, computer vision, information retrieval, and artificial intelligence. Each case study consists of two lectures: one an intuitive exposition of relevant computer-science concepts, and the other a hands-on introduction to a working system that embodies these concepts. Students use these systems to perform design and problem-solving tasks, thereby reinforcing the abstract concepts presented. Computer programming is neither required nor taught. The course has been offered for two years at the Harvard University Extension School, and has achieved high ratings in student surveys.

Graphical models, such as Bayesian networks and Markov random fields, represent statistical dependencies of variables by a graph. Local "belief propagation" rules of the sort proposed by Pearl (1988) are guaranteed to converge to the correct posterior probabilities in singly connected graphs. Recently, good performance has been obtained by using these same rules on graphs with loops, a method we refer to as loopy belief propagation. Perhaps the most dramatic instance is the near Shannon-limit performance of "Turbo codes," whose decoding algorithm is equivalent to loopy propagation. Except for the case of graphs with a single loop, there has been little theoretical understanding of loopy propagation. Here we analyze belief propagation in networks with arbitrary topologies when the nodes in the graph describe jointly gaussian random variables. We give an analytical formula relating the true posterior probabilities with those calculated using loopy propagation. We give sufficient conditions for convergence and show that when belief propagation converges, it gives the correct posterior means for all graph topologies, not just networks with a single loop. These results motivate using the powerful belief propagation algorithm in a broader class of networks and help clarify the empirical performance results.

Graphical models, such as Bayesian networks and Markov random fields (MRFs), represent statistical dependencies of variables by a graph. The max-product "belief propagation" algorithm is a local-message-passing algorithm on this graph that is known to converge to a unique fixed point when the graph is a tree. Furthermore, when the graph is a tree, the assignment based on the fixed point yields the most probable values of the unobserved variables given the observed ones. Good empirical performance has been obtained by running the max-product algorithm (or the equivalent min-sum algorithm) on graphs with loops, for applications including the decoding of "turbo" codes. Except for two simple graphs (cycle codes and single-loop graphs) there has been little theoretical understanding of the max-product algorithm on graphs with loops. Here we prove a result on the fixed points of max-product on a graph with arbitrary topology and with arbitrary probability distributions (discrete- or continuous-valued nodes). We show that the assignment based on a fixed point is a "neighborhood maximum" of the posterior probability: the posterior probability of the max-product assignment is guaranteed to be greater than all other assignments in a particular large region around that assignment. The region includes all assignments that differ from the max-product assignment in any subset of nodes that form no more than a single loop in the graph. In some graphs, this neighborhood is exponentially large. We illustrate the analysis with examples.

People can understand complex auditory and visual information, often using one to disambiguate the other. Automated analysis, even at a low(cid:173) level, faces severe challenges, including the lack of accurate statistical models for the signals, and their high-dimensionality and varied sam(cid:173) pling rates. Previous approaches [6] assumed simple parametric models for the joint distribution which, while tractable, cannot capture the com(cid:173) plex signal relationships. We learn the joint distribution of the visual and auditory signals using a non-parametric approach. First, we project the data into a maximally informative, low-dimensional subspace, suitable for density estimation. We then model the complicated stochastic rela(cid:173) tionships between the signals using a nonparametric density estimator. These learned densities allow processing across signal modalities. We demonstrate, on synthetic and real signals, localization in video of the face that is speaking in audio, and, conversely, audio enhancement of a particular speaker selected from the video.

Belief propagation (BP) was only supposed to work for tree-like networks but works surprisingly well in many applications involving networks with loops, including turbo codes. However, there has been little understanding of the algorithm or the nature of the solutions it finds for general graphs. We show that BP can only converge to a stationary point of an approximate free energy, known as the Bethe free energy in statis(cid:173) tical physics. This result characterizes BP fixed-points and makes connections with variational approaches to approximate inference. More importantly, our analysis lets us build on the progress made in statistical physics since Bethe's approximation was introduced in 1935. Kikuchi and others have shown how to construct more ac(cid:173) curate free energy approximations, of which Bethe's approximation is the simplest. Exploiting the insights from our analysis, we de(cid:173) rive generalized belief propagation (GBP) versions ofthese Kikuchi approximations. These new message passing algorithms can be significantly more accurate than ordinary BP, at an adjustable in(cid:173) crease in complexity. We illustrate such a new GBP algorithm on a grid Markov network and show that it gives much more accurate marginal probabilities than those found using ordinary BP.

Perceptual systems routinely separate "content" from "style," classifying familiar words spoken in an unfamiliar accent, identifying a font or handwriting style across letters, or recognizing a familiar face or object seen under unfamiliar viewing conditions. Yet a general and tractable computational model of this ability to untangle the underlying factors of perceptual observations remains elusive (Hofstadter, 1985). Existing factor models (Mardia, Kent, & Bibby, 1979; Hinton & Zemel, 1994; Ghahramani, 1995; Bell & Sejnowski, 1995; Hinton, Dayan, Frey, & Neal, 1995; Dayan, Hinton, Neal, & Zemel, 1995; Hinton & Ghahramani, 1997) are either insufficiently rich to capture the complex interactions of perceptually meaningful factors such as phoneme and speaker accent or letter and font, or do not allow efficient learning algorithms. We present a general framework for learning to solve two-factor tasks using bilinear models, which provide sufficiently expressive representations of factor interactions but can nonetheless be fit to data using efficient algorithms based on the singular value decomposition and expectation-maximization. We report promising results on three different tasks in three different perceptual domains: spoken vowel classification with a benchmark multi-speaker database, extrapolation of fonts to unseen letters, and translation of faces to novel illuminants.

We show a learning-based method for low-level vision problems-estimating scenes from images. We generate a synthetic world of scenes and their corresponding rendered images. We model that world with a Markov network, learning the network parameters from the examples. Bayesian belief propagation allows us to efficiently find a local maximum of the posterior probability for the scene, given the image. We call this approach VISTA-Vision by Image/Scene TrAining. We apply VISTA to the "super-resolution" problem (estimating high frequency details from a low-resolution image), showing good results. For the motion estimation problem, we show figure/ground discrimination, solution of the aperture problem, and filling-in arising from application of the same probabilistic machinery.

In many vision problems, we want to infer two (or more) hidden factors which interact to produce our observations. We may want to disentangle illuminant and object colors in color constancy; rendering conditions from surface shape in shape-from-shading; face identity and head pose in face recognition; or font and letter class in character recognition. We refer to these two factors generically as "style" and "content". Bilinear models offer a powerful framework for extracting the two-factor structure of a set of observations, and are familiar in computational vision from several well-known lines of research. This paper shows how bilinear models can be used to learn the style-content structure of a pattern analysis or synthesis problem, which can then be generalized to solve related tasks using different styles and/or content. We focus on three tasks: extrapolating the style of data to unseen content classes, classifying data with known content under a novel style, and translating data from novel content classes and style to a known style or content. We show examples from color constancy, face pose estimation, shape-from-shading, typography and speech.

Article Design galleries: a general approach to setting parameters for computer graphics and animation Share on Authors: J. Marks MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MA MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MAView Profile , B. Andalman Harvard Univ. Harvard Univ.View Profile , P. A. Beardsley MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MA MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MAView Profile , W. Freeman MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MA MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MAView Profile , S. Gibson MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MA MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MAView Profile , J. Hodgins Georgia Tech. Georgia Tech.View Profile , T. Kang CMU CMUView Profile , B. Mirtich MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MA MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MAView Profile , H. Pfister MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MA MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MAView Profile , W. Ruml MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MA MERL - A Mitsubishi Electric Research Laboratory, 201 Broadway, Cambridge, MAView Profile , K. Ryall Harvard Univ. Harvard Univ.View Profile , J. Seims Univ. of Washington Univ. of WashingtonView Profile , S. Shieber Harvard Univ. Harvard Univ.View Profile Authors Info & Claims SIGGRAPH '97: Proceedings of the 24th annual conference on Computer graphics and interactive techniquesAugust 1997 Pages 389–400https://doi.org/10.1145/258734.258887Online:03 August 1997Publication History 346citation2,992DownloadsMetricsTotal Citations346Total Downloads2,992Last 12 Months156Last 6 weeks16 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteGet Access

The appeal of computer games may be enhanced by vision-based user inputs. The high speed and low cost requirements for near-term, mass-market game applications make system design challenging. The response time of the vision interface should be less than a video frame time and the interface should cost less than $50 U.S. We meet these constraints with algorithms tailored to particular hardware. We have developed a special detector, called the artificial retina chip, which allows for fast, on-chip image processing. We describe two algorithms, based on image moments and orientation histograms, which exploit the capabilities of the chip to provide interactive response to the player's hand or body positions at 10 msec frame time and at low-cost. We show several possible game interactions.

We want to estimate the pose of human heads. This estimation involves a nonlinear mapping from the input image to an output parametric description. We characterize the mapping through examples from a training set, outputting the pose of the nearest example neighbor of the input. This is vector quantization, with the modification that we store an output parameter code with each quantized input code. For efficient indexing, we use a tree-structured vector quantizer (TSVQ). We make design choices based on the example application of monitoring an automobile driver's face. The reliance on stored data over computation power allows the system to be simple; efficient organization of the data allows it to be fast. We incorporate tracking in position and scale within the same vector quantization framework with virtually no cost in added computation. We show reasonable experimental results for a real-time prototype running on an inexpensive workstation.

Vision algorithms are often developed in a Bayesian framework. Two estimators are commonly used: maximum a posteriori (MAP), and minimum mean squared error (MMSE). We argue that neither is appropriate for perception problems. The MAP estimator makes insufficient use of structure in the posterior probability. The squared error penalty of the MMSE estimator does not reflect typical penalties. We describe a new estimator, which we call maximum local mass (MLM) [10, 26, 65], which integrates the local probability density. The MLM method is sensitive to local structure of the posterior probability, which MAP is not. The new method uses an optimality criterion that is appropriate for perception tasks: it finds the most probable approximately correct answer. For the case of low observation noise, we provide an efficient approximation. We apply this new estimator to color constancy. An unknown illuminant falls on surfaces of unknown colors. We seek to estimate both the illuminant spectrum and the surface spectra from photosensor responses which depend on the product of the unknown spectra. In simulations, we show that the MLM method performs better than the MAP estimator, and better than two standard color constancy algorithms. The MLM method may prove useful in other vision problems as well.>

We describe an architecture for efficient and accurate linear decomposition of an image into scale and orientation subbands. The basis functions of this decomposition are directional derivative operators of any desired order. We describe the construction and implementation of the transform.

The generic view assumption states that an observer is not in a special position relative to the scene. The author shows how to use the generic view assumption to quantify the likelihood of a view and so derive a powerful tool to estimate scene parameters. Generic variables can include viewpoint, object orientation, and lighting position. By considering the image as a function of these variables, the probability that a set of scene parameters created a given image is derived. This scene probability equation includes a term which rewards fidelity to the image, and a generic view term, which favors scenes likely to produce the observed image. Large derivatives of the image with respect to the generic variables correspond to unlikely scenes. This approach reduces the dependence on prior assumptions and should increase the scope and accuracy of scene estimates. This framework applies to many vision problems. Several shape from shading examples are shown, including the estimation of reflectance function, light direction, and vertical scale, in cases where these are otherwise unknown.>

One of the major drawbacks of orthogonal wavelet transforms is their lack of translation invariance: the content of wavelet subbands is unstable under translations of the input signal. Wavelet transforms are also unstable with respect to dilations of the input signal and, in two dimensions, rotations of the input signal. The authors formalize these problems by defining a type of translation invariance called shiftability. In the spatial domain, shiftability corresponds to a lack of aliasing; thus, the conditions under which the property holds are specified by the sampling theorem. Shiftability may also be applied in the context of other domains, particularly orientation and scale. Jointly shiftable transforms that are simultaneously shiftable in more than one domain are explored. Two examples of jointly shiftable transforms are designed and implemented: a 1-D transform that is jointly shiftable in position and scale, and a 2-D transform that is jointly shiftable in position and orientation. The usefulness of these image representations for scale-space analysis, stereo disparity measurement, and image enhancement is demonstrated.>

We describe a technique for displaying patterns that appear to move continuously without changing their positions. The method uses a quadrature pair of oriented filters to vary the local phase, giving the sensation of motion. We have used this technique in various computer graphic and scientific visualization applications.

The authors present an efficient architecture to synthesize filters of arbitrary orientations from linear combinations of basis filters, allowing one to adaptively steer a filter to any orientation, and to determine analytically the filter output as a function of orientation. Steerable filters may be designed in quadrature pairs to allow adaptive control over phase as well as orientation. The authors show how to design and steer the filters and present examples of their use in the analysis of orientation and phase, angularly adaptive filtering, edge detection, and shape from shading. One can also build a self-similar steerable pyramid representation. The same concepts can be generalized to the design of 3-D steerable filters.>

An efficient architecture is presented to synthesize filters of arbitrary orientations from linear combinations of basis filters, allowing one to adaptively 'steer' a filter to any orientation, and to determine analytically the filter output as a function of orientation. The authors show how to design and steer filters, and present examples of their use in several tasks: the analysis of orientation and phase, angularly adaptive filtering, edge detection, and shape-from-shading. It is also possible to build a self-similar steerable pyramid representation which may be considered to be a steerable wavelet transform. The same concepts can be generalized to the design of 3-D steerable filters, which should be useful in the analysis of image sequences and volumetric data.>