Multi-shot generation requires preserving the identity of characters and settings across frames. Cinematic scene composition goes beyond standard multi-shot generation, introducing additional challenges such as expressing complex interactions among multiple characters and visual effects to convey creative narratives—challenges existing datasets cannot fully address. We present CineVerse, a large-scale dataset of diverse movie scenes labeled with shot-level annotations tailored for filmmaking. CineVerse includes refined scene descriptions, shot-type information, and newly extracted shot, character, setting descriptions. We validate our dataset by developing a baseline framework that first generates a scene plan containing detailed information for the overall scene and each individual shot, then produces a set of coherent keyframes. Our results show significant improvements in controlling and synthesizing cinematic content through the added context provided by CineVerse.

Efficient neural representations for dynamic video scenes are critical for applications ranging from video compression to interactive simulations. Yet, existing methods often face challenges related to high memory usage, lengthy training times, and temporal consistency. To address these issues, we introduce a novel neural video representation that combines 3D Gaussian splatting with continuous camera motion modeling. By leveraging Neural ODEs, our approach learns smooth camera trajectories while maintaining an explicit 3D scene representation through Gaussians. Additionally, we introduce a spatiotemporal hierarchical learning strategy, progressively refining spatial and temporal features to enhance reconstruction quality and accelerate convergence. This memory-efficient approach achieves high-quality rendering at impressive speeds. Experimental results show that our hierarchical learning, combined with robust camera motion modeling, captures complex dynamic scenes with strong temporal consistency, achieving state-of-the-art performance across diverse video datasets in both high- and low-motion scenarios.

We present a novel perspective on learning video embedders for generative modeling: rather than requiring an exact reproduction of an input video, an effective embedder should focus on synthesizing visually plausible reconstructions. This relaxed criterion enables substantial improvements in compression ratios without compromising the quality of downstream generative models. Specifically, we propose replacing the conventional encoder-decoder video embedder with an encoder-generator framework that employs a diffusion transformer (DiT) to synthesize missing details from a compact latent space. Therein, we develop a dedicated latent conditioning module to condition the DiT decoder on the encoded video latent embedding. Our experiments demonstrate that our approach enables superior encoding-decoding performance compared to state-of-the-art methods, particularly as the compression ratio increases. To demonstrate the efficacy of our approach, we report results from our video embedders achieving a temporal compression ratio of up to 32x (8x higher than leading video embedders) and validate the robustness of this ultra-compact latent space for text-to-video generation, providing a significant efficiency boost in latent diffusion model training and inference.

Detecting object-level changes between two images across possibly different views (Fig. 1) is a core task in many applications that involve visual inspection or camera surveillance. Existing change-detection approaches suffer from three major limitations: (1) lack of evaluation on image pairs that contain no changes, leading to unreported false positive rates; (2) lack of correspondences (i.e., localizing the regions before and after a change); and (3) poor zero-shot generalization across different domains. To address these issues, we introduce a novel method that lever-ages change correspondences (a) during training to improve change detection accuracy, and (b) at test time, to minimize false positives. That is, we harness the supervision labels of where an object is added or removed to supervise change detectors, improving their accuracy over previous work [25] by a large margin. Our work is also the first to predict correspondences between pairs of detected changes using estimated homography and the Hungarian algorithm. Our model demonstrates superior performance over existing methods, achieving state-of-the-art results in change detection and change correspondence accuracy across both in-distribution and zero-shot benchmarks.

The integration of natural language processing (NLP) technologies into educational applications has shown promising results, particularly in the language learning domain. Many spoken open-domain chatbots have been used as speaking partners, helping language learners improve their language skills. However, one of the significant challenges is the high word-error-rate (WER) when recognising non-native/non-fluent speech, which interrupts conversation flow and leads to disappointment for learners. This paper explores the use of GPT4 for ASR error correction in conversational settings. In addition to WER, we propose to use semantic textual similarity (STS) and next response sensibility (NRS) metrics to evaluate the impact of correction models on conversation smoothness. We find that transcriptions corrected by GPT4 lead to higher conversation smoothness, despite an increase in WER. GPT4 also outperforms standard error correction methods without the need for in-domain training data.

We introduce MVDream, a diffusion model that is able to generate consistent multi-view images from a given text prompt. Learning from both 2D and 3D data, a multi-view diffusion model can achieve the generalizability of 2D diffusion models and the consistency of 3D renderings. We demonstrate that such a multi-view diffusion model is implicitly a generalizable 3D prior agnostic to 3D representations. It can be applied to 3D generation via Score Distillation Sampling, significantly enhancing the consistency and stability of existing 2D-lifting methods. It can also learn new concepts from a few 2D examples, akin to DreamBooth, but for 3D generation.

Implicit neural representation (INR) has been successful in representing static images. Contemporary image-based INR, with the use of Fourier-based positional encoding, can be viewed as a mapping from sinusoidal patterns with different frequencies to image content. Inspired by that view, we hypothesize that it is possible to generate temporally varying content with a single image-based INR model by displacing its input sinusoidal patterns over time. By exploiting the relation between the phase information in sinusoidal functions and their displacements, we incorporate into the conventional image-based INR model a phase-varying positional encoding module, and couple it with a phase-shift generation module that determines the phase-shift values at each frame. The model is trained end-to-end on a video to jointly determine the phase-shift values at each time with the mapping from the phase-shifted sinusoidal functions to the corresponding frame, enabling an implicit video representation. Experiments on a wide range of videos suggest that such a model is capable of learning to interpret phase-varying positional embeddings into the corresponding time-varying content. More importantly, we found that the learned phase-shift vectors tend to capture meaningful temporal and motion information from the video. In particular, manipulating the phase-shift vectors induces meaningful changes in the temporal dynamics of the resulting video, enabling non-trivial temporal and motion editing effects such as temporal interpolation, motion magnification, motion smoothing, and video loop detection.

The transcription quality of automatic speech recognition (ASR) systems degrades significantly when transcribing audios coming from unseen domains.We propose an unsupervised error correction method for unsupervised ASR domain adaption, aiming to recover transcription errors caused by domain mismatch.Unlike existing correction methods that rely on transcribed audios for training, our approach requires only unlabeled data of the target domains in which a pseudo-labeling technique is applied to generate correction training samples.To reduce over-fitting to the pseudo data, we also propose an encoder-decoder correction model that can take into account additional information such as dialogue context and acoustic features.Experiment results show that our method obtains a significant word error rate (WER) reduction over non-adapted ASR systems.The correction model can also be applied on top of other adaptation approaches to bring an additional improvement of 10% relatively.

Despite significant progress in monocular depth estimation in the wild, recent state-of-the-art methods cannot be used to recover accurate 3D scene shape due to an unknown depth shift induced by shift-invariant reconstruction losses used in mixed-data depth prediction training, and possible unknown camera focal length. We investigate this problem in detail, and propose a two-stage framework that first predicts depth up to an unknown scale and shift from a single monocular image, and then use 3D point cloud encoders to predict the missing depth shift and focal length that allow us to recover a realistic 3D scene shape. In addition, we propose an image-level normalized regression loss and a normal-based geometry loss to enhance depth prediction models trained on mixed datasets. We test our depth model on nine unseen datasets and achieve state-of-the-art performance on zero-shot dataset generalization. Code is available at: https://git.io/Depth

Neural networks have shown great abilities in estimating depth from a single image. However, the inferred depth maps are well below one-megapixel resolution and often lack fine-grained details, which limits their practicality. Our method builds on our analysis on how the input resolution and the scene structure affects depth estimation performance. We demonstrate that there is a trade-off between a consistent scene structure and the high-frequency details, and merge low- and high-resolution estimations to take advantage of this duality using a simple depth merging network. We present a double estimation method that improves the whole-image depth estimation and a patch selection method that adds local details to the final result. We demonstrate that by merging estimations at different resolutions with changing context, we can generate multi-megapixel depth maps with a high level of detail using a pre-trained model.

We present an algorithm for searching image collections using free-hand sketches that describe the appearance and relative positions of multiple objects1Sketch based image retrieval (SBIR) methods predominantly match queries containing a single, dominant object invariant to its position within an image. Our work exploits drawings as a concise and intuitive representation for specifying entire scene compositions. We train a convolutional neural network (CNN) to encode masked visual features from sketched objects, pooling these into a spatial descriptor encoding the spatial relationships and appearances of objects in the composition. Training the CNN backbone as a Siamese network under triplet loss yields a metric search embedding for measuring compositional similarity which may be efficiently leveraged for visual search by applying product quantization.

Existing deep learning based interactive segmentation methods have achieved remarkable performance with only a few user clicks, e.g. DEXTR [32] attaining 91.5% IoU on PASCAL VOC with only four extreme clicks. However, we observe even the state-of-the-art methods would often struggle in cases of objects to be segmented with elongated thin structures (e.g. bug legs and bicycle spokes). We investigate such failures, and find the critical reasons behind are two-fold: 1) lack of appropriate training dataset; and 2) extremely imbalanced distribution w.r.t. number of pixels belonging to thin and non-thin regions. Targeted at these challenges, we collect a large-scale dataset specifically for segmentation of thin elongated objects, named ThinObject-5K. Also, we present a novel integrative thin object segmentation network consisting of three streams. Among them, the high-resolution edge stream aims at preserving fine-grained details including elongated thin parts; the fixed-resolution context stream focuses on capturing semantic contexts. The two streams' outputs are then amalgamated in the fusion stream to complement each other for help producing a refined segmentation output with sharper predictions around thin parts. Extensive experimental results well demonstrate the effectiveness of our proposed solution on segmenting thin objects, surpassing the baseline by ~ 30% IoUthindespite using only four clicks. Codes and dataset are available at https://github.com/liewjunhao/thin-object-selection.

Video frame interpolation, the synthesis of novel views in time, is an increasingly popular research direction with many new papers further advancing the state of the art. But as each new method comes with a host of variables that affect the interpolation quality, it can be hard to tell what is actually important for this task. In this work, we show, somewhat surprisingly, that it is possible to achieve near state-of-the-art results with an older, simpler approach, namely adaptive separable convolutions, by a subtle set of low level improvements. In doing so, we propose a number of intuitive but effective techniques to improve the frame interpolation quality, which also have the potential to other related applications of adaptive convolutions such as burst image denoising, joint image filtering, or video prediction.

Current methods for active speaker detection focus on modeling audiovisual information from a single speaker. This strategy can be adequate for addressing single-speaker scenarios, but it prevents accurate detection when the task is to identify who of many candidate speakers are talking. This paper introduces the Active Speaker Context, a novel representation that models relationships between multiple speakers over long time horizons. Our new model learns pairwise and temporal relations from a structured ensemble of audiovisual observations. Our experiments show that a structured feature ensemble already benefits active speaker detection performance. We also find that the proposed Active Speaker Context improves the state-of-the-art on the AVA-ActiveSpeaker dataset achieving an mAP of 87.1%. Moreover, ablation studies verify that this result is a direct consequence of our long-term multi-speaker analysis.

While image captioning has progressed rapidly, existing works focus mainly on describing single images. In this paper, we introduce a new task, context-aware group captioning, which aims to describe a group of target images in the context of another group of related reference images. Context-aware group captioning requires not only summarizing information from both the target and reference image group but also contrasting between them. To solve this problem, we propose a framework combining self-attention mechanism with contrastive feature construction to effectively summarize common information from each image group while capturing discriminative information between them. To build the dataset for this task, we propose to group the images and generate the group captions based on single image captions using scene graphs matching. Our datasets are constructed on top of the public Conceptual Captions dataset and our new Stock Captions dataset. Experiments on the two datasets show the effectiveness of our method on this new task.

Single image depth prediction is a challenging task due to its ill-posed nature and challenges with capturing ground truth for supervision. Large-scale disparity data generated from stereo photos and 3D videos is a promising source of supervision, however, such disparity data can only approximate the inverse ground truth depth up to an affine transformation. To more effectively learn from such pseudo-depth data, we propose to use a simple pair-wise ranking loss with a novel sampling strategy. Instead of randomly sampling point pairs, we guide the sampling to better characterize structure of important regions based on the low-level edge maps and high-level object instance masks. We show that the pair-wise ranking loss, combined with our structure-guided sampling strategies, can significantly improve the quality of depth map prediction. In addition, we introduce a new relative depth dataset of about 21K diverse high-resolution web stereo photos to enhance the generalization ability of our model. In experiments, we conduct cross-dataset evaluation on six benchmark datasets and show that our method consistently improves over the baselines, leading to superior quantitative and qualitative results.

We present BlockGAN, an image generative model that learns object-aware 3D scene representations directly from unlabelled 2D images. Current work on scene representation learning either ignores scene background or treats the whole scene as one object. Meanwhile, work that considers scene compositionality treats scene objects only as image patches or 2D layers with alpha maps. Inspired by the computer graphics pipeline, we design BlockGAN to learn to first generate 3D features of background and foreground objects, then combine them into 3D features for the whole scene, and finally render them into realistic images. This allows BlockGAN to reason over occlusion and interaction between objects’ appearance, such as shadow and lighting, and provides control over each object’s 3D pose and identity, while maintaining image realism. BlockGAN is trained end-to-end, using only unlabelled single images, without the need for 3D geometry, pose labels, object masks, or multiple views of the same scene. Our experiments show that using explicit 3D features to represent objects allows BlockGAN to learn disentangled representations both in terms of objects (foreground and background) and their properties (pose and identity).

Abstract We present RGB2AO, a novel task to generate ambient occlusion (AO) from a single RGB image instead of screen space buffers such as depth and normal. RGB2AO produces a new image filter that creates a non‐directional shading effect that darkens enclosed and sheltered areas. RGB2AO aims to enhance two 2D image editing applications: image composition and geometry‐aware contrast enhancement. We first collect a synthetic dataset consisting of pairs of RGB images and AO maps. Subsequently, we propose a model for RGB2AO by supervised learning of a convolutional neural network (CNN), considering 3D geometry of the input image. Experimental results quantitatively and qualitatively demonstrate the effectiveness of our model.

Despite excellent performance on stationary test sets, deep neural networks (DNNs) can fail to generalize to out-of-distribution (OoD) inputs, including natural, non-adversarial ones, which are common in real-world settings. In this paper, we present a framework for discovering DNN failures that harnesses 3D renderers and 3D models. That is, we estimate the parameters of a 3D renderer that cause a target DNN to misbehave in response to the rendered image. Using our framework and a self-assembled dataset of 3D objects, we investigate the vulnerability of DNNs to OoD poses of well-known objects in ImageNet. For objects that are readily recognized by DNNs in their canonical poses, DNNs incorrectly classify 97% of their pose space. In addition, DNNs are highly sensitive to slight pose perturbations. Importantly, adversarial poses transfer across models and datasets. We find that 99.9% and 99.4% of the poses misclassified by Inception-v3 also transfer to the AlexNet and ResNet-50 image classifiers trained on the same ImageNet dataset, respectively, and 75.5% transfer to the YOLOv3 object detector trained on MS COCO.

Existing deep learning-based interactive image segmentation approaches typically assume the target-of-interest is always a single object and fail to account for the potential diversity in user expectations, thus requiring excessive user input when it comes to segmenting an object part or a group of objects instead. Motivated by the observation that the object part, full object, and a collection of objects essentially differ in size, we propose a new concept called scale-diversity, which characterizes the spectrum of segmentations w.r.t. different scales. To address this, we present MultiSeg, a scale-diverse interactive image segmentation network that incorporates a set of two-dimensional scale priors into the model to generate a set of scale-varying proposals that conform to the user input. We explicitly encourage segmentation diversity during training by synthesizing diverse training samples for a given image. As a result, our method allows the user to quickly locate the closest segmentation target for further refinement if necessary. Despite its simplicity, experimental results demonstrate that our proposed model is capable of quickly producing diverse yet plausible segmentation outputs, reducing the user interaction required, especially in cases where many types of segmentations (object parts or groups) are expected.

We propose a novel video inpainting algorithm that simultaneously hallucinates missing appearance and motion (optical flow) information, building upon the recent 'Deep Image Prior' (DIP) that exploits convolutional network architectures to enforce plausible texture in static images. In extending DIP to video we make two important contributions. First, we show that coherent video inpainting is possible without a priori training. We take a generative approach to inpainting based on internal (within-video) learning without reliance upon an external corpus of visual data to train a one-size-fits-all model for the large space of general videos. Second, we show that such a framework can jointly generate both appearance and flow, whilst exploiting these complementary modalities to ensure mutual consistency. We show that leveraging appearance statistics specific to each video achieves visually plausible results whilst handling the challenging problem of long-term consistency.

The Ken Burns effect allows animating still images with a virtual camera scan and zoom. Adding parallax, which results in the 3D Ken Burns effect, enables significantly more compelling results. Creating such effects manually is time-consuming and demands sophisticated editing skills. Existing automatic methods, however, require multiple input images from varying viewpoints. In this paper, we introduce a framework that synthesizes the 3D Ken Burns effect from a single image, supporting both a fully automatic mode and an interactive mode with the user controlling the camera. Our framework first leverages a depth prediction pipeline, which estimates scene depth that is suitable for view synthesis tasks. To address the limitations of existing depth estimation methods such as geometric distortions, semantic distortions, and inaccurate depth boundaries, we develop a semantic-aware neural network for depth prediction, couple its estimate with a segmentation-based depth adjustment process, and employ a refinement neural network that facilitates accurate depth predictions at object boundaries. According to this depth estimate, our framework then maps the input image to a point cloud and synthesizes the resulting video frames by rendering the point cloud from the corresponding camera positions. To address disocclusions while maintaining geometrically and temporally coherent synthesis results, we utilize context-aware color- and depth-inpainting to fill in the missing information in the extreme views of the camera path, thus extending the scene geometry of the point cloud. Experiments with a wide variety of image content show that our method enables realistic synthesis results. Our study demonstrates that our system allows users to achieve better results while requiring little effort compared to existing solutions for the 3D Ken Burns effect creation.

The performance of image retrieval has been improved tremendously in recent years through the use of deep feature representations. Most existing methods, however, aim to retrieve images that are visually similar or semantically relevant to the query, irrespective of spatial configuration. In this paper, we develop a spatial-semantic image search technology that enables users to search for images with both semantic and spatial constraints by manipulating concept text-boxes on a 2D query canvas. We train a convolutional neural network to synthesize appropriate visual features that captures the spatial-semantic constraints from the user canvas query. We directly optimize the retrieval performance of the visual features when training our deep neural network. These visual features then are used to retrieve images that are both spatially and semantically relevant to the user query. The experiments on large-scale datasets such as MS-COCO and Visual Genome show that our method outperforms other baseline and state-of-the-art methods in spatial-semantic image search.

Video frame interpolation typically involves two steps: motion estimation and pixel synthesis. Such a two-step approach heavily depends on the quality of motion estimation. This paper presents a robust video frame interpolation method that combines these two steps into a single process. Specifically, our method considers pixel synthesis for the interpolated frame as local convolution over two input frames. The convolution kernel captures both the local motion between the input frames and the coefficients for pixel synthesis. Our method employs a deep fully convolutional neural network to estimate a spatially-adaptive convolution kernel for each pixel. This deep neural network can be directly trained end to end using widely available video data without any difficult-to-obtain ground-truth data like optical flow. Our experiments show that the formulation of video interpolation as a single convolution process allows our method to gracefully handle challenges like occlusion, blur, and abrupt brightness change and enables high-quality video frame interpolation.

Standard video frame interpolation methods first estimate optical flow between input frames and then synthesize an intermediate frame guided by motion. Recent approaches merge these two steps into a single convolution process by convolving input frames with spatially adaptive kernels that account for motion and re-sampling simultaneously. These methods require large kernels to handle large motion, which limits the number of pixels whose kernels can be estimated at once due to the large memory demand. To address this problem, this paper formulates frame interpolation as local separable convolution over input frames using pairs of 1D kernels. Compared to regular 2D kernels, the 1D kernels require significantly fewer parameters to be estimated. Our method develops a deep fully convolutional neural network that takes two input frames and estimates pairs of 1D kernels for all pixels simultaneously. Since our method is able to estimate kernels and synthesizes the whole video frame at once, it allows for the incorporation of perceptual loss to train the neural network to produce visually pleasing frames. This deep neural network is trained end-to-end using widely available video data without any human annotation. Both qualitative and quantitative experiments show that our method provides a practical solution to high-quality video frame interpolation.

Photo aesthetics assessment is challenging. Deep convolutional neural network (ConvNet) methods have recently shown promising results for aesthetics assessment. The performance of these deep ConvNet methods, however, is often compromised by the constraint that the neural network only takes the fixed-size input. To accommodate this requirement, input images need to be transformed via cropping, scaling, or padding, which often damages image composition, reduces image resolution, or causes image distortion, thus compromising the aesthetics of the original images. In this paper, we present a composition-preserving deep Con-vNet method that directly learns aesthetics features from the original input images without any image transformations. Specifically, our method adds an adaptive spatial pooling layer upon the regular convolution and pooling layers to directly handle input images with original sizes and aspect ratios. To allow for multi-scale feature extraction, we develop the Multi-Net Adaptive Spatial Pooling ConvNet architecture which consists of multiple sub-networks with different adaptive spatial pooling sizes and leverage a scene-based aggregation layer to effectively combine the predictions from multiple sub-networks. Our experiments on the large-scale aesthetics assessment benchmark (AVA [29]) demonstrate that our method can significantly improve the state-of-the-art results in photo aesthetics assessment.

Kernel estimation for image deblurring is a challenging task and a large number of algorithms have been developed. Our hypothesis is that while individual kernels estimated using different methods alone are sometimes inadequate, they often complement each other. This paper addresses the problem of fusing multiple kernels estimated using different methods into a more accurate one that can better support image deblurring than each individual kernel. In this paper, we develop a data-driven approach to kernel fusion that learns how each kernel contributes to the final kernel and how they interact with each other. We discuss various kernel fusion models and find that kernel fusion using Gaussian Conditional Random Fields performs best. This Gaussian Conditional Random Fields-based kernel fusion method not only models how individual kernels are fused at each kernel element but also the interaction of kernel fusion among multiple kernel elements. Our experiments show that our method can significantly improve image deblurring by combining kernels from multiple methods into a better one.

A variety of methods have been developed for visual saliency analysis. These methods often complement each other. This paper addresses the problem of aggregating various saliency analysis methods such that the aggregation result outperforms each individual one. We have two major observations. First, different methods perform differently in saliency analysis. Second, the performance of a saliency analysis method varies with individual images. Our idea is to use data-driven approaches to saliency aggregation that appropriately consider the performance gaps among individual methods and the performance dependence of each method on individual images. This paper discusses various data-driven approaches and finds that the image-dependent aggregation method works best. Specifically, our method uses a Conditional Random Field (CRF) framework for saliency aggregation that not only models the contribution from individual saliency map but also the interaction between neighboring pixels. To account for the dependence of aggregation on an individual image, our approach selects a subset of images similar to the input image from a training data set and trains the CRF aggregation model only using this subset instead of the whole training set. Our experiments on public saliency benchmarks show that our aggregation method outperforms each individual saliency method and is robust with the selection of aggregated methods.

Simplicity refers to one of the most important photography composition rules. Simplicity states that simplifying the image background can draw viewers' attention to the subject of interest in a photograph and help them better comprehend and appreciate it. Understanding whether a photo respects photography rules or not facilitates photo quality assessment. In this paper, we present a method to automatically detect whether a photo is composed according to the rule of simplicity. We design features according to the definition, implementation and effect of the rule. First, we make use of saliency analysis to infer the subject of interest in a photo and measure its compactness. Second, we segment an image into background and foreground and measure the homogeneity within the background as another feature. Third, when looking at an image created with the rule of simplicity, different viewers tend to agree on what the subject of interest is in this photo. We accordingly measure the consistency among various saliency detection results as a feature. We experiment with these features in a range of machine learning methods. Our experiments show that our methods, together with these features, provide an encouraging result in detecting the rule of simplicity in a photo.

The rule of thirds is one of the most important composition rules used by photographers to create high-quality photos. The rule of thirds states that placing important objects along the imagery thirds lines or around their intersections often produces highly aesthetic photos. In this paper, we present a method to automatically determine whether a photo respects the rule of thirds. Detecting the rule of thirds from a photo requires semantic content understanding to locate important objects, which is beyond the state of the art. This paper makes use of the recent saliency and generic objectness analysis as an alternative and accordingly designs a range of features. Our experiment with a variety of saliency and generic objectness methods shows that an encouraging performance can be achieved in detecting the rule of thirds from photos.