The spatio-temporal relationship between the pixels of a video carries critical information for low-level 4D perception tasks. A single model that reasons about it should be able to solve several such tasks well. Yet, most state-of-theart methods rely on architectures specialized for the task at hand. We present L4P, a feedforward, general-purpose architecture that solves low-level 4D perception tasks in a unified framework. LAP leverages a pre-trained ViT-based video encoder and combines it with per-task heads that are lightweight and therefore do not require extensive training. Despite its general and feedforward formulation, our method is competitive with existing specialized methods on both dense tasks, such as depth or optical flow estimation, and sparse tasks, such as 2D/3D tracking. Moreover, it solves all tasks at once in a time comparable to that of single-task methods.

Large models have shown generalization across datasets for many low-level vision tasks, like depth estimation, but no such general models exist for scene flow. Even though scene flow prediction has wide potential, its practical use is limited because of the lack of generalization of current predictive models. We identify three key challenges and propose solutions for each. First, we create a method that jointly estimates geometry and motion for accurate prediction. Second, we alleviate scene flow data scarcity with a data recipe that affords us 1M annotated training samples across diverse synthetic scenes. Third, we evaluate different parameterizations for scene flow prediction and adopt a natural and effective parameterization. Our model outperforms existing methods as well as baselines built on large-scale models in terms of 3D end-point error, and shows zero-shot generalization to the casually captured videos from DAVIS and the robotic manipulation scenes from RoboTAP. Overall, our approach makes scene flow prediction more practical in-the-wild.Website: https://research.nvidia.com/labs/lpr/zeromsf/

Tremendous progress has been made in deep stereo matching to excel on benchmark datasets through per-domain fine-tuning. However, achieving strong zero-shot generalization — a hallmark of foundation models in other computer vision tasks — remains challenging for stereo matching. We introduce FoundationStereo, a foundation model for stereo depth estimation designed to achieve strong zero-shot generalization. To this end, we first construct a large-scale (1M stereo pairs) synthetic training dataset featuring large diversity and high photorealism, followed by an automatic self-curation pipeline to remove ambiguous samples. We then design a number of network architecture components to enhance scalability, including a side-tuning feature backbone that adapts rich monocular priors from vision foundation models to mitigate the sim-to-real gap, and long-range context reasoning for effective cost volume filtering. Together, these components lead to strong robustness and accuracy across domains, establishing a new standard in zero-shot stereo depth estimation. Project page: https://nvlabs.github.io/FoundationStereo/

AI-based talking-head videoconferencing systems reduce bandwidth by transmitting a latent representation of a speaker’s pose and expression, which is used to synthesize frames on the receiver's end. However, these systems are vulnerable to “puppeteering” attacks, where an adversary controls the identity of another person in real-time. Traditional deepfake detectors fail here, as all video content is synthetic. We propose a novel biometric defense that detects identity leakage in the transmitted latent representation. Our metric-learning approach disentangles identity cues from pose and expression, enabling detection of unauthorized swaps. Experiments across multiple talking-head models show that our method consistently outperforms prior defenses, operates in real time on consumer GPUs, and generalizes well to out-of-distribution data. By targeting the latent features shared during normal operation, our method offers a practical and robust safeguard against puppeteering.

Wide field-of-view (FoV) cameras efficiently capture large portions of the scene, which makes them attractive in multiple domains, such as automotive and robotics. For such applications, estimating depth from multiple images is a critical task, and therefore, a large amount of ground truth (GT) data is available. Unfortunately, most of the GT data is for pinhole cameras, making it impossible to properly train depth estimation models for large-FoV cameras. We propose the first method to train a stereo depth estimation model on the widely available pinhole data, and to generalize it to data captured with larger FoVs. Our intuition is simple: We warp the training data to a canonical, large-FoV representation and augment it to allow a single network to reason about diverse types of distortions that otherwise would prevent generalization. We show strong generalization ability of our approach on both indoor and outdoor datasets, which was not possible with previous methods.

Transferring the pose of a reference avatar to stylized 3D characters of various shapes is a fundamental task in computer graphics. Existing methods either require the stylized characters to be rigged, or they use the stylized character in the desired pose as ground truth at training. We present a zero-shot approach that requires only the widely available deformed non-stylized avatars in training, and deforms stylized characters of significantly different shapes at inference. Classical methods achieve strong generalization by deforming the mesh at the triangle level, but this requires labelled correspondences. We leverage the power of local deformation, but without requiring explicit correspondence labels. We introduce a semi-supervised shape-understanding module to bypass the need for explicit correspondences at test time, and an implicit pose deformation module that deforms individual surface points to match the target pose. Further-more, to encourage realistic and accurate deformation of stylized characters, we introduce an efficient volume-based test-time training procedure. Because it does not need rigging, nor the deformed stylized character at training time, our model generalizes to categories with scarce annotation, such as stylized quadrupeds. Extensive experiments demonstrate the effectiveness of the proposed method compared to the state-of-the-art approaches trained with comparable or more supervision. Our project page is available at https://jiashunwang.github.io/ZPT/

Unsupervised generation of high-quality multi-view-consistent images and 3D shapes using only collections of single-view 2D photographs has been a long-standing challenge. Existing 3D GANs are either compute intensive or make approximations that are not 3D-consistent; the former limits quality and resolution of the generated images and the latter adversely affects multi-view consistency and shape quality. In this work, we improve the computational efficiency and image quality of 3D GANs without overly relying on these approximations. We introduce an expressive hybrid explicit implicit network architecture that, together with other design choices, synthesizes not only high-resolution multi-view-consistent images in real time but also produces high-quality 3D geometry. By decoupling feature generation and neural rendering, our framework is able to leverage state-of-the-art 2D CNN generators, such as StyleGAN2, and inherit their efficiency and expressiveness. We demonstrate state-of-the-art 3D-aware synthesis with FFHQ and AFHQ Cats, among other experiments.

Rendering articulated objects while controlling their poses is critical to applications such as virtual reality or animation for movies. Manipulating the pose of an object, however, requires the understanding of its underlying structure, that is, its joints and how they interact with each other. Unfortunately, assuming the structure to be known, as existing methods do, precludes the ability to work on new object categories. We propose to learn both the appearance and the structure of previously unseen articulated objects by ob-serving them move from multiple views, with no joints annotation supervision, or information about the structure. We observe that 3D points that are static relative to one another should belong to the same part, and that adjacent parts that move relative to each other must be connected by a joint. To leverage this insight, we model the object parts in 3D as ellipsoids, which allows us to identify joints. We combine this explicit representation with an implicit one that compensates for the approximation introduced. We show that our method works for different structures, from quadrupeds, to single-arm robots, to humans. The code is available at https://github.com/NVlabs/watch-it-move and a version of this manuscript that uses animations is at https://arxiv.org/abs/2112.11347

Time-to-contact (TTC), the time for an object to collide with the observer’s plane, is a powerful tool for path planning: it is potentially more informative than the depth, velocity, and acceleration of objects in the scene—even for humans. TTC presents several advantages, including requiring only a monocular, uncalibrated camera. However, regressing TTC for each pixel is not straightforward, and most existing methods make over-simplifying assumptions about the scene. We address this challenge by estimating TTC via a series of simpler, binary classifications. We predict with low latency whether the observer will collide with an obstacle within a certain time, which is often more critical than knowing exact, per-pixel TTC. For such scenarios, our method offers a temporal geofence in 6.4 ms—over 25× faster than existing methods. Our approach can also estimate per-pixel TTC with arbitrarily fine quantization (including continuous values), when the computational budget allows for it. To the best of our knowledge, our method is the first to offer TTC information (binary or coarsely quantized) at sufficiently high frame-rates for practical use.

Estimating a mesh from an unordered set of sparse, noisy 3D points is a challenging problem that requires to carefully select priors. Existing hand-crafted priors, such as smoothness regularizers, impose an undesirable trade-off between attenuating noise and preserving local detail. Recent deep-learning approaches produce impressive results by learning priors directly from the data. However, the priors are learned at the object level, which makes these algorithms class-specific, and even sensitive to the pose of the object. We introduce meshlets, small patches of mesh that we use to learn local shape priors. Meshlets act as a dictionary of local features and thus allow to use learned priors to reconstruct object meshes in any pose and from unseen classes, even when the noise is large and the samples sparse.

Stereo-based depth estimation is a cornerstone of computer vision, with state-of-the-art methods delivering accurate results in real time. For several applications such as autonomous navigation, however, it may be useful to trade accuracy for lower latency. We present Bi3D, a method that estimates depth via a series of binary classifications. Rather than testing if objects are at a particular depth D, as existing stereo methods do, it classifies them as being closer or farther than D. This property offers a powerful mechanism to balance accuracy and latency. Given a strict time budget, Bi3D can detect objects closer than a given distance in as little as a few milliseconds, or estimate depth with arbitrarily coarse quantization, with complexity linear with the number of quantization levels. Bi3D can also use the allotted quantization levels to get continuous depth, but in a specific depth range. For standard stereo (i.e., continuous depth on the whole range), our method is close to or on par with state-of-the-art, finely tuned stereo methods.

This paper presents a new method to synthesize an image from arbitrary views and times given a collection of images of a dynamic scene. A key challenge for the novel view synthesis arises from dynamic scene reconstruction where epipolar geometry does not apply to the local motion of dynamic contents. To address this challenge, we propose to combine the depth from single view (DSV) and the depth from multi-view stereo (DMV), where DSV is complete, i.e., a depth is assigned to every pixel, yet view-variant in its scale, while DMV is view-invariant yet incomplete. Our insight is that although its scale and quality are inconsistent with other views, the depth estimation from a single view can be used to reason about the globally coherent geometry of dynamic contents. We cast this problem as learning to correct the scale of DSV, and to refine each depth with locally consistent motions between views to form a coherent depth estimation. We integrate these tasks into a depth fusion network in a self-supervised fashion. Given the fused depth maps, we synthesize a photorealistic virtual view in a specific location and time with our deep blending network that completes the scene and renders the virtual view. We evaluate our method of depth estimation and view synthesis on a diverse real-world dynamic scenes and show the outstanding performance over existing methods.

Content creation, central to applications such as virtual reality, can be tedious and time-consuming. Recent image synthesis methods simplify this task by offering tools to generate new views from as little as a single input image, or by converting a semantic map into a photorealistic image. We propose to push the envelope further, and introduce Generative View Synthesis (GVS) that can synthesize multiple photorealistic views of a scene given a single semantic map. We show that the sequential application of existing techniques, e.g., semantics-to-image translation followed by monocular view synthesis, fail at capturing the scene's structure. In contrast, we solve the semantics-to-image translation in concert with the estimation of the 3D layout of the scene, thus producing geometrically consistent novel views that preserve semantic structures. We first lift the input 2D semantic map onto a 3D layered representation of the scene in feature space, thereby preserving the semantic labels of 3D geometric structures. We then project the layered features onto the target views to generate the final novel-view images. We verify the strengths of our method and compare it with several advanced baselines on three different datasets. Our approach also allows for style manipulation and image editing operations, such as the addition or removal of objects, with simple manipulations of the input style images and semantic maps respectively. For code and additional results, visit the project page at https://gvsnet.github.io

Convolutions are the fundamental building blocks of CNNs. The fact that their weights are spatially shared is one of the main reasons for their widespread use, but it is also a major limitation, as it makes convolutions content-agnostic. We propose a pixel-adaptive convolution (PAC) operation, a simple yet effective modification of standard convolutions, in which the filter weights are multiplied with a spatially varying kernel that depends on learnable, local pixel features. PAC is a generalization of several popular filtering techniques and thus can be used for a wide range of use cases. Specifically, we demonstrate state-of-the-art performance when PAC is used for deep joint image upsampling. PAC also offers an effective alternative to fully-connected CRF (Full-CRF), called PAC-CRF, which performs competitively compared to Full-CRF, while being considerably faster. In addition, we also demonstrate that PAC can be used as a drop-in replacement for convolution layers in pre-trained networks, resulting in consistent performance improvements.

We present Extreme View Synthesis, a solution for novel view extrapolation that works even when the number of input images is small---as few as two. In this context, occlusions and depth uncertainty are two of the most pressing issues, and worsen as the degree of extrapolation increases. We follow the traditional paradigm of performing depth-based warping and refinement, with a few key improvements. First, we estimate a depth probability volume, rather than just a single depth value for each pixel of the novel view. This allows us to leverage depth uncertainty in challenging regions, such as depth discontinuities. After using it to get an initial estimate of the novel view, we explicitly combine learned image priors and the depth uncertainty to synthesize a refined image with less artifacts. Our method is the first to show visually pleasing results for baseline magnifications of up to 30x.

To date, top-performing optical flow estimation methods only take pairs of consecutive frames into account. While elegant and appealing, the idea of using more than two frames has not yet produced state-of-the-art results. We present a simple, yet effective fusion approach for multi-frame optical flow that benefits from longer-term temporal cues. Our method first warps the optical flow from previous frames to the current, thereby yielding multiple plausible estimates. It then fuses the complementary information carried by these estimates into a new optical flow field. At the time of writing, our method ranks first among published results in the MPI Sintel and KITTI 2015 benchmarks. Our models will be available on https://github.com/NVlabs/PWC-Net.

Motion blur is a fundamental problem in computer vision as it impacts image quality and hinders inference. Traditional deblurring algorithms leverage the physics of the image formation model and use hand-crafted priors: they usually produce results that better reflect the underlying scene, but present artifacts. Recent learning-based methods implicitly extract the distribution of natural images directly from the data and use it to synthesize plausible images. Their results are impressive, but they are not always faithful to the content of the latent image. We present an approach that bridges the two. Our method fine-tunes existing deblurring neural networks in a self-supervised fashion by enforcing that the output, when blurred based on the optical flow between subsequent frames, matches the input blurry image. We show that our method significantly improves the performance of existing methods on several datasets both visually and in terms of image quality metrics.

Capturing a picture that "tells a story" requires the ability to create the right composition. The two most important parameters controlling composition are the camera position and the focal length of the lens. The traditional paradigm is for a photographer to mentally visualize the desired picture, select the capture parameters to produce it, and finally take the photograph, thus committing to a particular composition. We propose to change this paradigm. To do this, we introduce computational zoom , a framework that allows a photographer to manipulate several aspects of composition in post-processing from a stack of pictures captured at different distances from the scene. We further define a multi-perspective camera model that can generate compositions that are not physically attainable, thus extending the photographer's control over factors such as the relative size of objects at different depths and the sense of depth of the picture. We show several applications and results of the proposed computational zoom framework.

The human visual system can operate in a wide range of illumination levels due to several adaptation processes working in concert. For the most part, these adaptation mechanisms are transparent, leaving the observer unaware of his or her absolute adaptation state. At extreme illumination levels, however, some of these mechanisms produce perceivable secondary effects, or epiphenomena. In bright light, these include bleaching afterimages and adaptation afterimages, while in dark conditions these include desaturation, loss of acuity, mesopic hue shift, and the Purkinje effect. In this work we examine whether displaying these effects explicitly can be used to extend the apparent dynamic range of a conventional computer display. We present phenomenological models for each effect, describe efficient computer graphics methods for rendering our models, and propose a gaze-adaptive display that injects the effects into imagery on a standard computer monitor. Finally, we report the results of psychophysical experiments which reveal that, while mesopic epiphenomena are a strong cue that a stimulus is very dark, afterimages have little impact on the perception that a stimulus is very bright.

We present a novel method for aligning images in an HDR (high-dynamic-range) image stack to produce a new exposure stack where all the images are aligned and appear as if they were taken simultaneously, even in the case of highly dynamic scenes. Our method produces plausible results even where the image used as a reference is either too dark or bright to allow for an accurate registration.

Abstract When creating a High‐Dynamic‐Range (HDR) image from a sequence of differently exposed Low‐Dynamic‐Range (LDR) images, the set of LDR images is usually generated by sampling the space of exposure times with a geometric progression and without explicitly accounting for the distribution of irradiance values of the scene. We argue that this choice can produce sub‐optimal results both in terms of the number of acquired pictures and the quality of the resulting HDR image. This paper presents a method to estimate the full irradiance histogram of a scene, and a strategy to select the set of exposures that need to be acquired. Our selection usually requires a smaller or equal set of LDRs, yet produces higher quality HDR images.

Camera cellphones have become ubiquitous, thus opening a plethora of opportunities for mobile vision applications. For instance, they can enable users to access reviews or price comparisons for a product from a picture of its barcode while still in the store. Barcode reading needs to be robust to challenging conditions such as blur, noise, low resolution, or low-quality camera lenses, all of which are extremely common. Surprisingly, even state-of-the-art barcode reading algorithms fail when some of these factors come into play. One reason resides in the early commitment strategy that virtually all existing algorithms adopt: The image is first binarized and then only the binary data are processed. We propose a new approach to barcode decoding that bypasses binarization. Our technique relies on deformable templates and exploits all of the gray-level information of each pixel. Due to our parameterization of these templates, we can efficiently perform maximum likelihood estimation independently on each digit and enforce spatial coherence in a subsequent step. We show by way of experiments on challenging UPC-A barcode images from five different databases that our approach outperforms competing algorithms. Implemented on a Nokia N95 phone, our algorithm can localize and decode a barcode on a VGA image (640 × 480, JPEG compressed) in an average time of 400-500 ms.

Current camera-based barcode readers do not work well when the image has low resolution, is out of focus, or is motion-blurred. One main reason is that virtually all existing algorithms perform some sort of binarization, either by gray scale thresholding or by finding the bar edges. We propose a new approach to barcode reading that never needs to binarize the image. Instead, we use deformable barcode digit models in a maximum likelihood setting. We show that the particular nature of these models enables efficient integration over the space of deformations. Global optimization over all digits is then performed using dynamic programming. Experiments with challenging UPC-A barcode images show substantial improvement over other state-of-the-art algorithms.

As the applications delivered by cellular phones are becoming increasingly sophisticated, the importance of choosing an input strategy is also growing. Touch-screens can simplify navigation by far but the vast majority of phones on the market are not equipped with them. Cameras, on the other hand, are widespread even amongst low-end phones: In this paper we propose a vision-based pointing system that allows the user to control the pointer's position by just waving a hand, with no need for additional hardware.