Search

Existing homography and optical flow methods are erroneous in challenging scenes, such as fog, rain, night, and snow because the basic assumptions such as brightness and gradient constancy are broken. To address this issue, we present an unsupervised learning approach that fuses gyroscope into homography and optical flow learning. Specifically, we first convert gyroscope readings into motion fields named gyro field. Second, we design a self-guided fusion module (SGF) to fuse the background motion extracted from the gyro field with the optical flow and guide the network to focus on motion details. Meanwhile, we propose a homography decoder module (HD) to combine gyro field and intermediate results of SGF to produce the homography. To the best of our knowledge, this is the first deep learning framework that fuses gyroscope data and image content for both deep homography and optical flow learning. To validate our method, we propose a new dataset that covers regular and challenging scenes. Experiments show that our method outperforms the state-of-the-art methods in both regular and challenging scenes. The code and dataset are available at https://github.com/lhaippp/GyroFlowPlus. [ABSTRACT FROM AUTHOR]

Hierarchical concepts have proven useful in many classical and learning-based optical flow methods regarding both accuracy and robustness. In this paper we show that such concepts are still useful in the context of recent neural networks that follow RAFT's paradigm refraining from hierarchical strategies by relying on recurrent updates based on a single-scale all-pairs transform. To this end, we introduce MS-RAFT+: a novel recurrent multi-scale architecture based on RAFT that unifies several successful hierarchical concepts. It employs a coarse-to-fine estimation to enable the use of finer resolutions by useful initializations from coarser scales. Moreover, it relies on RAFT's correlation pyramid that allows to consider non-local cost information during the matching process. Furthermore, it makes use of advanced multi-scale features that incorporate high-level information from coarser scales. And finally, our method is trained subject to a sample-wise robust multi-scale multi-iteration loss that closely supervises each iteration on each scale, while allowing to discard particularly difficult samples. In combination with an appropriate mixed-dataset training strategy, our method performs favorably. It not only yields highly accurate results on the four major benchmarks (KITTI 2015, MPI Sintel, Middlebury and VIPER), it also allows to achieve these results with a single model and a single parameter setting. Our trained model and code are available at https://github.com/cv-stuttgart/MS_RAFT_plus. [ABSTRACT FROM AUTHOR]

Recent optical flow estimators usually employ deep models designed for image classification as the encoders for feature extraction and matching. However, those encoders developed for image classification may be sub-optimal for flow estimation. In contrast, the decoder design of optical flow estimators often requires meticulous design for flow estimation. The disconnect between the encoder and decoder could negatively affect optical flow estimation. To address this issue, we propose a neural architecture search method, FlowNAS, to automatically find the more suitable and stronger encoder architecture for existing flow decoders. We first design a suitable search space, including various convolutional operators, and construct a weight-sharing super-network for efficiently evaluating the candidate architectures. To better train the super-network, we present a Feature Alignment Distillation module that utilizes a well-trained flow estimator to guide the training of the super-network. Finally, a resource-constrained evolutionary algorithm is exploited to determine an optimal architecture (i.e., sub-network). Experimental results show that FlowNAS can be easily incorporated into existing flow estimators and achieves state-of-the-art performance with the trade-off between accuracy and efficiency. Furthermore, the encoder architecture discovered by FlowNAS with the weights inherited from the super-network achieves 4.67% F1-all error on KITTI, an 8.4% reduction of RAFT baseline, surpassing state-of-the-art handcrafted GMA and AGFlow models, while reducing the model complexity and latency. The source code and trained models will be released at https://github.com/VDIGPKU/FlowNAS. [ABSTRACT FROM AUTHOR]

In this paper, we focus on exploring effective methods for faster and accurate semantic segmentation. A common practice to improve the performance is to attain high-resolution feature maps with strong semantic representation. Two strategies are widely used: atrous convolutions and feature pyramid fusion, while both are either computationally intensive or ineffective. Inspired by the Optical Flow for motion alignment between adjacent video frames, we propose a Flow Alignment Module (FAM) to learn Semantic Flow between feature maps of adjacent levels and broadcast high-level features to high-resolution features effectively and efficiently. Furthermore, integrating our FAM to a standard feature pyramid structure exhibits superior performance over other real-time methods, even on lightweight backbone networks, such as ResNet-18 and DFNet. Then to further speed up the inference procedure, we also present a novel Gated Dual Flow Alignment Module to directly align high-resolution feature maps and low-resolution feature maps where we term the improved version network as SFNet-Lite. Extensive experiments are conducted on several challenging datasets, where results show the effectiveness of both SFNet and SFNet-Lite. In particular, when using Cityscapes test set, the SFNet-Lite series achieve 80.1 mIoU while running at 60 FPS using ResNet-18 backbone and 78.8 mIoU while running at 120 FPS using STDC backbone on RTX-3090. Moreover, we unify four challenging driving datasets (i.e., Cityscapes, Mapillary, IDD, and BDD) into one large dataset, which we named Unified Driving Segmentation (UDS) dataset. It contains diverse domain and style information. We benchmark several representative works on UDS. Both SFNet and SFNet-Lite still achieve the best speed and accuracy trade-off on UDS, which serves as a strong baseline in such a challenging setting. The code and models are publicly available at https://github.com/lxtGH/SFSegNets. [ABSTRACT FROM AUTHOR]

A good understanding of geometrical concepts as well as a broad familiarity with objects lead to excellent human perception of moving objects. The human ability to detect and segment moving objects works in the presence of multiple objects, complex background geometry, motion of the observer and even camouflage. How we perceive moving objects so reliably is a longstanding research question in computer vision and borrows findings from related areas such as psychology, cognitive science and physics. One approach to the problem is to teach a deep network to model all of these effects. This is in contrast with the strategy used by human vision, where cognitive processes and body design are tightly coupled and each is responsible for certain aspects of correctly identifying moving objects. Similarly, from the computer vision perspective there is evidence that classical, geometry-based techniques are better suited to the "motion-based" parts of the problem, while deep networks are more suitable for modeling appearance. In this work, we argue that the coupling of camera rotation and camera translation can create complex motion fields that are difficult for a deep network to untangle directly. We present a novel probabilistic model to estimate the camera's rotation given the motion field. We then rectify the flow field to obtain a rotation-compensated motion field for subsequent segmentation. This strategy of first estimating camera motion, and then allowing a network to learn the remaining parts of the problem, yields improved results on the widely used DAVIS benchmark as well as the more recent motion segmentation data set MoCA (Moving Camouflaged Animals). [ABSTRACT FROM AUTHOR]

A popular approach to estimate scene flow is to utilize point cloud data from various Lidar scans. However, there is little attention to learning 3D motion from camera images. Learning scene flow from a monocular camera remains a challenging task due to its ill-posedness as well as lack of annotated data. Self-supervised methods demonstrate learning scene flow estimation from unlabeled data, yet their accuracy lags behind (semi-)supervised methods. In this paper, we introduce a self-supervised monocular scene flow method that substantially improves the accuracy over the previous approaches. Based on RAFT, a state-of-the-art optical flow model, we design a new decoder to iteratively update 3D motion fields and disparity maps simultaneously. Furthermore, we propose an enhanced upsampling layer and a disparity initialization technique, which overall further improves accuracy up to 7.2%. Our method achieves state-of-the-art accuracy among all self-supervised monocular scene flow methods, improving accuracy by 34.2%. Our fine-tuned model outperforms the best previous semi-supervised method with 228 times faster runtime. Code will be publicly available to ensure reproducibility. [ABSTRACT FROM AUTHOR]

Advances in visual perceptual tasks have been mainly driven by the amount, and types, of annotations of large-scale datasets. Researchers have focused on fully-supervised settings to train models using offline epoch-based schemes. Despite the evident advancements, limitations and cost of manually annotated datasets have hindered further development for event perceptual tasks, such as detection and localization of objects and events in videos. The problem is more apparent in zoological applications due to the scarcity of annotations and length of videos-most videos are at most ten minutes long. Inspired by cognitive theories, we present a self-supervised perceptual prediction framework to tackle the problem of temporal event segmentation by building a stable representation of event-related objects. The approach is simple but effective. We rely on LSTM predictions of high-level features computed by a standard deep learning backbone. For spatial segmentation, the stable representation of the object is used by an attention mechanism to filter the input features before the prediction step. The self-learned attention maps effectively localize the object as a side effect of perceptual prediction. We demonstrate our approach on long videos from continuous wildlife video monitoring, spanning multiple days at 25 FPS. We aim to facilitate automated ethogramming by detecting and localizing events without the need for labels. Our approach is trained in an online manner on streaming input and requires only a single pass through the video, with no separate training set. Given the lack of long and realistic (includes real-world challenges) datasets, we introduce a new wildlife video dataset–nest monitoring of the Kagu (a flightless bird from New Caledonia)–to benchmark our approach. Our dataset features a video from 10 days (over 23 million frames) of continuous monitoring of the Kagu in its natural habitat. We annotate every frame with bounding boxes and event labels. Additionally, each frame is annotated with time-of-day and illumination conditions. We will make the dataset, which is the first of its kind, and the code available to the research community. We find that the approach significantly outperforms other self-supervised, traditional (e.g., Optical Flow, Background Subtraction) and NN-based (e.g., PA-DPC, DINO, iBOT), baselines and performs on par with supervised boundary detection approaches (i.e., PC). At a recall rate of 80%, our best performing model detects one false positive activity every 50 min of training. On average, we at least double the performance of self-supervised approaches for spatial segmentation. Additionally, we show that our approach is robust to various environmental conditions (e.g., moving shadows). We also benchmark the framework on other datasets (i.e., Kinetics-GEBD, TAPOS) from different domains to demonstrate its generalizability. The data and code are available on our project page: https://aix.eng.usf.edu/research_automated_ethogramming.html [ABSTRACT FROM AUTHOR]

Birds of prey rely on vision to execute flight manoeuvres that are key to their survival, such as intercepting fast-moving targets or navigating through clutter. A better understanding of the role played by vision during these manoeuvres is not only relevant within the field of animal behaviour, but could also have applications for autonomous drones. In this paper, we present a novel method that uses computer vision tools to analyse the role of active vision in bird flight, and demonstrate its use to answer behavioural questions. Combining motion capture data from Harris' hawks with a hybrid 3D model of the environment, we render RGB images, semantic maps, depth information and optic flow outputs that characterise the visual experience of the bird in flight. In contrast with previous approaches, our method allows us to consider different camera models and alternative gaze strategies for the purposes of hypothesis testing, allows us to consider visual input over the complete visual field of the bird, and is not limited by the technical specifications and performance of a head-mounted camera light enough to attach to a bird's head in flight. We present pilot data from three sample flights: a pursuit flight, in which a hawk intercepts a moving target, and two obstacle avoidance flights. With this approach, we provide a reproducible method that facilitates the collection of large volumes of data across many individuals, opening up new avenues for data-driven models of animal behaviour. [ABSTRACT FROM AUTHOR]

Video frame interpolation aims to synthesize new intermediate frames between existing ones, which is an important task in video enhancement. A classic direction in this field is flow-based which estimates motions in the form of optical flow, warps the frames, and synthesizes the final results. In this work, we explicitly investigate the warping step and propose a way to combine the strength from using both forward and backward warping. Our method, named HWFI, introduces hybrid warping fusion for frame interpolation. We also include edge information explicitly in our pipeline and employ channel attention in our synthesis network. Compared to the latest state-of-the-art method that only uses forward warping, our method produces better results with higher quality, especially in edge regions. Extensive experiments show that our method can obtain the best results qualitatively and quantitatively on multiple benchmark datasets. [ABSTRACT FROM AUTHOR]

Unsupervised optical flow and stereo depth estimation are two fundamental tasks in computer vision. Current studies (Tosi et al., in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp 4654–4665, 2020; Ranjan et al., in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp 12240–12249, 2019; Wang et al., in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp 8071–8081, 2019; Yin and Shi, in: Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 1983–1992, 2018) demonstrate that jointly learning networks for optical flow and stereo depth estimation via the geometric constraints can mutually benefit the two tasks and in turn yield large accuracy improvements. However, most of these methods generate geometric constraints based on estimated camera pose, which are not applicable to scenarios with moving objects that have different motions from the camera. In addition, errors of estimated camera pose would yield inaccurate constraints for the two tasks. In this paper, we propose a novel and universal geometric loss function, named Subspace-PnP, which is based on the Perspective-n-Points (PnP) and union-of-subspaces theory (Ji et al., in: IEEE Winter conference on applications of computer vision, pp 461–468, 2014) to jointly estimate the optical flow and stereo depth. The construction of Subspace-PnP dose not rely on the camera pose, but implicitly contains information of camera pose and motions of all moving objects. Our experiments show that the Subspace-PnP loss can mutually guide the estimation of optical flow and depth, enabling better robustness and greater accuracy even in dynamic scenes. In addition, we propose a motion-occlusion simulation method to handle occlusions caused by moving objects in optical flow estimation, which in turn can yield further performance improvement. Our method achieves the state-of-the-art performance for joint optical flow and stereo depth estimation on the KITTI 2012 and KITTI 2015 benchmarks. [ABSTRACT FROM AUTHOR]

We introduce the first approach to solve the challenging problem of automatic 4D visual scene understanding for complex dynamic scenes with multiple interacting people from multi-view video. Our approach simultaneously estimates a detailed model that includes a per-pixel semantically and temporally coherent reconstruction, together with instance-level segmentation exploiting photo-consistency, semantic and motion information. We further leverage recent advances in 3D pose estimation to constrain the joint semantic instance segmentation and 4D temporally coherent reconstruction. This enables per person semantic instance segmentation of multiple interacting people in complex dynamic scenes. Extensive evaluation of the joint visual scene understanding framework against state-of-the-art methods on challenging indoor and outdoor sequences demonstrates a significant ( ≈ 40 % ) improvement in semantic segmentation, reconstruction and scene flow accuracy. In addition to the evaluation on several indoor and outdoor scenes, the proposed joint 4D scene understanding framework is applied to challenging outdoor sports scenes in the wild captured with manually operated wide-baseline broadcast cameras. [ABSTRACT FROM AUTHOR]

Ghosting artifacts caused by moving objects and misalignments are a key challenge in constructing high dynamic range (HDR) images. Current methods first register the input low dynamic range (LDR) images using optical flow before merging them. This process is error-prone, and often causes ghosting in the resulting merged image. We propose a novel dual-attention-guided end-to-end deep neural network, called DAHDRNet, which produces high-quality ghost-free HDR images. Unlike previous methods that directly stack the LDR images or features for merging, we use dual-attention modules to guide the merging according to the reference image. DAHDRNet thus exploits both spatial attention and feature channel attention to achieve ghost-free merging. The spatial attention modules automatically suppress undesired components caused by misalignments and saturation, and enhance the fine details in the non-reference images. The channel attention modules adaptively rescale channel-wise features by considering the inter-dependencies between channels. The dual-attention approach is applied recurrently to further improve feature representation, and thus alignment. A dilated residual dense block is devised to make full use of the hierarchical features and increase the receptive field when hallucinating missing details. We employ a hybrid loss function, which consists of a perceptual loss, a total variation loss, and a content loss to recover photo-realistic images. Although DAHDRNet is not flow-based, it can be applied to flow-based registration to reduce artifacts caused by optical-flow estimation errors. Experiments on different datasets show that the proposed DAHDRNet achieves state-of-the-art quantitative and qualitative results. [ABSTRACT FROM AUTHOR]

Dual-view snapshot compressive imaging (SCI) aims to capture videos from two field-of-views (FoVs) using a 2D sensor (detector) in a single snapshot, achieving joint FoV and temporal compressive sensing, and thus enjoying the advantages of low-bandwidth, low-power and low-cost. However, it is challenging for existing model-based decoding algorithms to reconstruct each individual scene, which usually require exhaustive parameter tuning with extremely long running time for large scale data. In this paper, we propose an optical flow-aided recurrent neural network for dual video SCI systems, which provides high-quality decoding in seconds. Firstly, we develop a diversity amplification method to enlarge the differences between scenes of two FoVs, and design a deep convolutional neural network with dual branches to separate different scenes from the single measurement. Secondly, we integrate the bidirectional optical flow extracted from adjacent frames with the recurrent neural network to jointly reconstruct each video in a sequential manner. Extensive results on both simulation and real data demonstrate the superior performance of our proposed model in short inference time. The code and data are available at https://github.com/RuiyingLu/OFaNet-for-Dual-view-SCI. [ABSTRACT FROM AUTHOR]

Event cameras have numerous advantages over traditional cameras, such as low latency, high temporal resolution, and high dynamic range (HDR). We initially investigate the potential of creating intensity images/videos from an adjustable portion of the event data stream via event-based conditional generative adversarial networks (cGANs). Using the proposed framework, we further show the versatility of our method in directly handling similar supervised tasks, such as optical flow and depth prediction. Stacks of space-time coordinates of events are used as the inputs while the proposed framework is trained to predict either the intensity images, optical flows, or depth outputs according to the target task. We further demonstrate the unique capability of our approach in generating HDR images even under extreme illumination conditions, creating non-blurred images under rapid motion, and generating very high frame rate videos up to the temporal resolution of event cameras. The proposed framework is evaluated using a publicly available real-world dataset and a synthetic dataset we prepared by utilizing an event camera simulator. [ABSTRACT FROM AUTHOR]

Face images captured through glass are usually contaminated by reflections. The low-transmitted reflections make the reflection removal more challenging than for general scenes because important facial features would be completely occluded. In this paper, we propose and solve the face image reflection removal problem. We recover the important facial structures by incorporating inpainting ideas into a guided reflection removal framework, which takes two images as the input and considers various face-specific priors. We use a newly collected face reflection image dataset to train our model and compare with state-of-the-art methods. The proposed method shows advantages in estimating reflection-free face images for improving face recognition. [ABSTRACT FROM AUTHOR]

Deep video action recognition models have been highly successful in recent years but require large quantities of manually-annotated data, which are expensive and laborious to obtain. In this work, we investigate the generation of synthetic training data for video action recognition, as synthetic data have been successfully used to supervise models for a variety of other computer vision tasks. We propose an interpretable parametric generative model of human action videos that relies on procedural generation, physics models and other components of modern game engines. With this model we generate a diverse, realistic, and physically plausible dataset of human action videos, called PHAV for "Procedural Human Action Videos". PHAV contains a total of 39,982 videos, with more than 1000 examples for each of 35 action categories. Our video generation approach is not limited to existing motion capture sequences: 14 of these 35 categories are procedurally-defined synthetic actions. In addition, each video is represented with 6 different data modalities, including RGB, optical flow and pixel-level semantic labels. These modalities are generated almost simultaneously using the Multiple Render Targets feature of modern GPUs. In order to leverage PHAV, we introduce a deep multi-task (i.e. that considers action classes from multiple datasets) representation learning architecture that is able to simultaneously learn from synthetic and real video datasets, even when their action categories differ. Our experiments on the UCF-101 and HMDB-51 benchmarks suggest that combining our large set of synthetic videos with small real-world datasets can boost recognition performance. Our approach also significantly outperforms video representations produced by fine-tuning state-of-the-art unsupervised generative models of videos. [ABSTRACT FROM AUTHOR]

The optical flow of humans is well known to be useful for the analysis of human action. Recent optical flow methods focus on training deep networks to approach the problem. However, the training data used by them does not cover the domain of human motion. Therefore, we develop a dataset of multi-human optical flow and train optical flow networks on this dataset. We use a 3D model of the human body and motion capture data to synthesize realistic flow fields in both single- and multi-person images. We then train optical flow networks to estimate human flow fields from pairs of images. We demonstrate that our trained networks are more accurate than a wide range of top methods on held-out test data and that they can generalize well to real image sequences. The code, trained models and the dataset are available for research. [ABSTRACT FROM AUTHOR]

Convolutional neural networks (CNNs) handle the case where filters extend beyond the image boundary using several heuristics, such as zero, repeat or mean padding. These schemes are applied in an ad-hoc fashion and, being weakly related to the image content and oblivious of the target task, result in low output quality at the boundary. In this paper, we propose a simple and effective improvement that learns the boundary handling itself. At training-time, the network is provided with a separate set of explicit boundary filters. At testing-time, we use these filters which have learned to extrapolate features at the boundary in an optimal way for the specific task. Our extensive evaluation, over a wide range of architectural changes (variations of layers, feature channels, or both), shows how the explicit filters result in improved boundary handling. Furthermore, we investigate the efficacy of variations of such boundary filters with respect to convergence speed and accuracy. Finally, we demonstrate an improvement of 5–20% across the board of typical CNN applications (colorization, de-Bayering, optical flow, disparity estimation, and super-resolution). Supplementary material and code can be downloaded from the project page: http://geometry.cs.ucl.ac.uk/projects/2019/investigating-edge/. [ABSTRACT FROM AUTHOR]

This paper presents novel techniques for recovering 3D dense scene flow, based on differential analysis of 4D light fields. The key enabling result is a per-ray linear equation, called the ray flow equation, that relates 3D scene flow to 4D light field gradients. The ray flow equation is invariant to 3D scene structure and applicable to a general class of scenes, but is under-constrained (3 unknowns per equation). Thus, additional constraints must be imposed to recover motion. We develop two families of scene flow algorithms by leveraging the structural similarity between ray flow and optical flow equations: local 'Lucas–Kanade' ray flow and global 'Horn–Schunck' ray flow, inspired by corresponding optical flow methods. We also develop a combined local–global method by utilizing the correspondence structure in the light fields. We demonstrate high precision 3D scene flow recovery for a wide range of scenarios, including rotation and non-rigid motion. We analyze the theoretical and practical performance limits of the proposed techniques via the light field structure tensor, a 3 × 3 matrix that encodes the local structure of light fields. We envision that the proposed analysis and algorithms will lead to design of future light-field cameras that are optimized for motion sensing, in addition to depth sensing. [ABSTRACT FROM AUTHOR]

Many video enhancement algorithms rely on optical flow to register frames in a video sequence. Precise flow estimation is however intractable; and optical flow itself is often a sub-optimal representation for particular video processing tasks. In this paper, we propose task-oriented flow (TOFlow), a motion representation learned in a self-supervised, task-specific manner. We design a neural network with a trainable motion estimation component and a video processing component, and train them jointly to learn the task-oriented flow. For evaluation, we build Vimeo-90K, a large-scale, high-quality video dataset for low-level video processing. TOFlow outperforms traditional optical flow on standard benchmarks as well as our Vimeo-90K dataset in three video processing tasks: frame interpolation, video denoising/deblocking, and video super-resolution. [ABSTRACT FROM AUTHOR]

We present the focal flow sensor. It is an unactuated, monocular camera that simultaneously exploits defocus and differential motion to measure a depth map and a 3D scene velocity field. It does this using an optical-flow-like, per-pixel linear constraint that relates image derivatives to depth and velocity. We derive this constraint, prove its invariance to scene texture, and prove that it is exactly satisfied only when the sensor’s blur kernels are Gaussian. We analyze the inherent sensitivity of the focal flow cue, and we build and test a prototype. Experiments produce useful depth and velocity information for a broader set of aperture configurations, including a simple lens with a pillbox aperture. [ABSTRACT FROM AUTHOR]

Previous joint/guided filters directly transfer structural information from the reference to the target image. In this paper, we analyze the major drawback-that is, there may be completely different edges in the two images. Simply considering all patterns could introduce significant errors. To address this issue, we propose the concept of mutual-structure, which refers to the structural information that is contained in both images and thus can be safely enhanced by joint filtering. We also use an untraditional objective function that can be efficiently optimized to yield mutual structure. Our method results in important edge preserving property, which greatly benefits depth completion, optical flow estimation, image enhancement, stereo matching, to name a few. [ABSTRACT FROM AUTHOR]

We introduce a novel matching algorithm, called DeepMatching, to compute dense correspondences between images. DeepMatching relies on a hierarchical, multi-layer, correlational architecture designed for matching images and was inspired by deep convolutional approaches. The proposed matching algorithm can handle non-rigid deformations and repetitive textures and efficiently determines dense correspondences in the presence of significant changes between images. We evaluate the performance of DeepMatching, in comparison with state-of-the-art matching algorithms, on the Mikolajczyk (Mikolajczyk et al. A comparison of affine region detectors, 2005), the MPI-Sintel (Butler et al. A naturalistic open source movie for optical flow evaluation, 2012) and the Kitti (Geiger et al. Vision meets robotics: The KITTI dataset, 2013) datasets. DeepMatching outperforms the state-of-the-art algorithms and shows excellent results in particular for repetitive textures. We also apply DeepMatching to the computation of optical flow, called DeepFlow, by integrating it in the large displacement optical flow (LDOF) approach of Brox and Malik (Large displacement optical flow: descriptor matching in variational motion estimation, 2011). Additional robustness to large displacements and complex motion is obtained thanks to our matching approach. DeepFlow obtains competitive performance on public benchmarks for optical flow estimation. [ABSTRACT FROM AUTHOR]

This paper introduces a state-of-the-art video representation and applies it to efficient action recognition and detection. We first propose to improve the popular dense trajectory features by explicit camera motion estimation. More specifically, we extract feature point matches between frames using SURF descriptors and dense optical flow. The matches are used to estimate a homography with RANSAC. To improve the robustness of homography estimation, a human detector is employed to remove outlier matches from the human body as human motion is not constrained by the camera. Trajectories consistent with the homography are considered as due to camera motion, and thus removed. We also use the homography to cancel out camera motion from the optical flow. This results in significant improvement on motion-based HOF and MBH descriptors. We further explore the recent Fisher vector as an alternative feature encoding approach to the standard bag-of-words (BOW) histogram, and consider different ways to include spatial layout information in these encodings. We present a large and varied set of evaluations, considering (i) classification of short basic actions on six datasets, (ii) localization of such actions in feature-length movies, and (iii) large-scale recognition of complex events. We find that our improved trajectory features significantly outperform previous dense trajectories, and that Fisher vectors are superior to BOW encodings for video recognition tasks. In all three tasks, we show substantial improvements over the state-of-the-art results. [ABSTRACT FROM AUTHOR]

3D scene flow estimation aims to jointly recover dense geometry and 3D motion from stereoscopic image sequences, thus generalizes classical disparity and 2D optical flow estimation. To realize its conceptual benefits and overcome limitations of many existing methods, we propose to represent the dynamic scene as a collection of rigidly moving planes, into which the input images are segmented. Geometry and 3D motion are then jointly recovered alongside an over-segmentation of the scene. This piecewise rigid scene model is significantly more parsimonious than conventional pixel-based representations, yet retains the ability to represent real-world scenes with independent object motion. It, furthermore, enables us to define suitable scene priors, perform occlusion reasoning, and leverage discrete optimization schemes toward stable and accurate results. Assuming the rigid motion to persist approximately over time additionally enables us to incorporate multiple frames into the inference. To that end, each view holds its own representation, which is encouraged to be consistent across all other viewpoints and frames in a temporal window. We show that such a view-consistent multi-frame scheme significantly improves accuracy, especially in the presence of occlusions, and increases robustness against adverse imaging conditions. Our method currently achieves leading performance on the KITTI benchmark, for both flow and stereo. [ABSTRACT FROM AUTHOR]

Invariances are one of the key concepts to render computer vision algorithms robust against severe illumination changes. However, there is no free lunch: With any invariance comes an unavoidable loss of information. The goal of our paper is to introduce two novel descriptors which minimise this loss: the complete rank transform and the complete census transform. They are invariant under monotonically increasing intensity rescalings, while containing a maximum possible amount of information. To analyse our descriptors, we embed them as constancy assumptions into a variational framework for optic flow computation. As a suitable regularisation term, we choose total generalised variation that favours piecewise affine solutions. Our experiments focus on the KITTI benchmark where robustness w.r.t. illumination changes is one of the main issues. The results demonstrate that our descriptors yield state-of-the-art accuracy. [ABSTRACT FROM AUTHOR]

In the context of extracting information from video, bad weather conditions like rain can have a detrimental effect. In this paper, a novel framework to detect and remove rain streaks from video is proposed. The first part of the proposed framework for rain removal is a technique to detect rain streaks based on phase congruency features. The variation of features from frame to frame is used to estimate the candidate rain pixels in a frame. In order to reduce the number of false candidates due to global motion, frames are registered using phase correlation. The second part of the proposed framework is a novel reconstruction technique that utilizes information from three different sources, which are intensities of the rain affected pixel, spatial neighbors, and temporal neighbors. An optimal estimate for the actual intensity of the rain affected pixel is made based on the minimization of registration error between frames. An optical flow technique using local phase information is adopted for registration. This part of the proposed framework for removing rain is modeled such that the presence of local motion will not distort the features in the reconstructed video. The proposed framework is evaluated quantitatively and qualitatively on a variety of videos with varying complexities. The effectiveness of the algorithm is quantitatively verified by computing a no-reference image quality measure on individual frames of the reconstructed video. From a variety of experiments that are performed on output videos, it is shown that the proposed technique performs better than state-of-the-art techniques. [ABSTRACT FROM AUTHOR]

We present a method for producing dense active appearance models (AAMs), suitable for video-realistic synthesis. To this end we estimate a joint alignment of all training images using a set of pairwise registrations and ensure that these pairwise registrations are only calculated between similar images. This is achieved by defining a graph on the image set whose edge weights correspond to registration errors and computing a bounded diameter minimum spanning tree. Dense optical flow is used to compute pairwise registration and a flow refinement method to align small scale texture is introduced. Further, given the registration of training images, vertices are added to the AAM to minimise the error between the observed flow fields and the flow fields interpolated between the AAM mesh points. We demonstrate a significant improvement in model compactness. [ABSTRACT FROM AUTHOR]

The accuracy of optical flow estimation algorithms has been improving steadily as evidenced by results on the Middlebury optical flow benchmark. The typical formulation, however, has changed little since the work of Horn and Schunck. We attempt to uncover what has made recent advances possible through a thorough analysis of how the objective function, the optimization method, and modern implementation practices influence accuracy. We discover that 'classical' flow formulations perform surprisingly well when combined with modern optimization and implementation techniques. One key implementation detail is the median filtering of intermediate flow fields during optimization. While this improves the robustness of classical methods it actually leads to higher energy solutions, meaning that these methods are not optimizing the original objective function. To understand the principles behind this phenomenon, we derive a new objective function that formalizes the median filtering heuristic. This objective function includes a non-local smoothness term that robustly integrates flow estimates over large spatial neighborhoods. By modifying this new term to include information about flow and image boundaries we develop a method that can better preserve motion details. To take advantage of the trend towards video in wide-screen format, we further introduce an asymmetric pyramid downsampling scheme that enables the estimation of longer range horizontal motions. The methods are evaluated on the Middlebury, MPI Sintel, and KITTI datasets using the same parameter settings. [ABSTRACT FROM AUTHOR]

This paper addresses the problem of non-rigid video registration, or the computation of optical flow from a reference frame to each of the subsequent images in a sequence, when the camera views deformable objects. We exploit the high correlation between 2D trajectories of different points on the same non-rigid surface by assuming that the displacement of any point throughout the sequence can be expressed in a compact way as a linear combination of a low-rank motion basis. This subspace constraint effectively acts as a trajectory regularization term leading to temporally consistent optical flow. We formulate it as a robust soft constraint within a variational framework by penalizing flow fields that lie outside the low-rank manifold. The resulting energy functional can be decoupled into the optimization of the brightness constancy and spatial regularization terms, leading to an efficient optimization scheme. Additionally, we propose a novel optimization scheme for the case of vector valued images, based on the dualization of the data term. This allows us to extend our approach to deal with colour images which results in significant improvements on the registration results. Finally, we provide a new benchmark dataset, based on motion capture data of a flag waving in the wind, with dense ground truth optical flow for evaluation of multi-frame optical flow algorithms for non-rigid surfaces. Our experiments show that our proposed approach outperforms state of the art optical flow and dense non-rigid registration algorithms. [ABSTRACT FROM AUTHOR]

In this work we present a gradient-based variational model for video editing, addressing the problem of propagating gradient-domain information along the optical flow of the video. The resulting propagation is temporally consistent and blends seamlessly with its spatial surroundings. In addition, the presented model is able to cope with additive illumination changes and handles occlusions/dis-occlusions. The problem of propagation along the optical flow arises in different video editing applications. In this work we consider the application where a user edits a frame by modifying the texture of an object's surface and wishes to propagate this editing throughout the video. [ABSTRACT FROM AUTHOR]

This paper introduces a video representation based on dense trajectories and motion boundary descriptors. Trajectories capture the local motion information of the video. A dense representation guarantees a good coverage of foreground motion as well as of the surrounding context. A state-of-the-art optical flow algorithm enables a robust and efficient extraction of dense trajectories. As descriptors we extract features aligned with the trajectories to characterize shape (point coordinates), appearance (histograms of oriented gradients) and motion (histograms of optical flow). Additionally, we introduce a descriptor based on motion boundary histograms (MBH) which rely on differential optical flow. The MBH descriptor shows to consistently outperform other state-of-the-art descriptors, in particular on real-world videos that contain a significant amount of camera motion. We evaluate our video representation in the context of action classification on nine datasets, namely KTH, YouTube, Hollywood2, UCF sports, IXMAS, UIUC, Olympic Sports, UCF50 and HMDB51. On all datasets our approach outperforms current state-of-the-art results. [ABSTRACT FROM AUTHOR]

We propose a depth and image scene flow estimation method taking the input of a binocular video. The key component is motion-depth temporal consistency preservation, making computation in long sequences reliable. We tackle a number of fundamental technical issues, including connection establishment between motion and depth, structure consistency preservation in multiple frames, and long-range temporal constraint employment for error correction. We address all of them in a unified depth and scene flow estimation framework. Our main contributions include development of motion trajectories, which robustly link frame correspondences in a voting manner, rejection of depth/motion outliers through temporal robust regression, novel edge occurrence map estimation, and introduction of anisotropic smoothing priors for proper regularization. [ABSTRACT FROM AUTHOR]

The tracking of deformable objects using video data is a demanding research topic due to the inherent ambiguity problems, which can only be solved using additional assumptions about the deformation. Image feature points, commonly used to approach the deformation problem, only provide sparse information about the scene at hand. In this paper a tracking approach for deformable objects in color and depth video is introduced that does not rely on feature points or optical flow data but employs all the input image information available to find a suitable deformation for the data at hand. A versatile NURBS based deformation space is defined for arbitrary complex triangle meshes, decoupling the object surface complexity from the complexity of the deformation. An efficient optimization scheme is introduced that is able to calculate results in real-time (25 Hz). Extensive synthetic and real data tests of the algorithm and its features show the reliability of this approach. [ABSTRACT FROM AUTHOR]

We tackle the problem of detecting occluded regions in a video stream. Under assumptions of Lambertian reflection and static illumination, the task can be posed as a variational optimization problem, and its solution approximated using convex minimization. We describe efficient numerical schemes that reach the global optimum of the relaxed cost functional, for any number of independently moving objects, and any number of occlusion layers. We test the proposed algorithm on benchmark datasets, expanded to enable evaluation of occlusion detection performance, in addition to optical flow. [ABSTRACT FROM AUTHOR]

This paper deals with plane detection from a monocular image sequence without camera calibration or a priori knowledge about the egomotion. Within a framework of driver assistance applications, it is assumed that the 3D scene is a set of 3D planes. In this paper, the vision process considers obstacles, roads and buildings as planar structures. These planes are detected by exploiting iso-velocity curves after optical flow estimation. A Hough Transform-like frame called c- velocity was designed. This paper explains how this c- velocity, defined by analogy to the v- disparity in stereovision, can represent planes, regardless of their orientation and how this representation facilitates plane extraction. Under a translational camera motion, planar surfaces are transformed into specific parabolas of the c- velocity space. The error and robustness analysis of the proposed technique confirms that this cumulative approach is very efficient for making the detection more robust and coping with optical flow imprecision. Moreover, the results suggest that the concept could be generalized to the detection of other parameterized surfaces than planes. [ABSTRACT FROM AUTHOR]

There are two main strategies for solving correspondence problems in computer vision: sparse local feature based approaches and dense global energy based methods. While sparse feature based methods are often used for estimating the fundamental matrix by matching a small set of sophistically optimised interest points, dense energy based methods mark the state of the art in optical flow computation. The goal of our paper is to show that this separation into different application domains is unnecessary and can be bridged in a natural way. As a first contribution we present a new application of dense optical flow for estimating the fundamental matrix. Comparing our results with those obtained by feature based techniques we identify cases in which dense methods have advantages over sparse approaches. Motivated by these promising results we propose, as a second contribution, a new variational model that recovers the fundamental matrix and the optical flow simultaneously as the minimisers of a single energy functional. In experiments we show that our coupled approach is able to further improve the estimates of both the fundamental matrix and the optical flow. Our results prove that dense variational methods can be a serious alternative even in classical application domains of sparse feature based approaches. [ABSTRACT FROM AUTHOR]

The quantitative evaluation of optical flow algorithms by Barron et al. () led to significant advances in performance. The challenges for optical flow algorithms today go beyond the datasets and evaluation methods proposed in that paper. Instead, they center on problems associated with complex natural scenes, including nonrigid motion, real sensor noise, and motion discontinuities. We propose a new set of benchmarks and evaluation methods for the next generation of optical flow algorithms. To that end, we contribute four types of data to test different aspects of optical flow algorithms: (1) sequences with nonrigid motion where the ground-truth flow is determined by tracking hidden fluorescent texture, (2) realistic synthetic sequences, (3) high frame-rate video used to study interpolation error, and (4) modified stereo sequences of static scenes. In addition to the average angular error used by Barron et al., we compute the absolute flow endpoint error, measures for frame interpolation error, improved statistics, and results at motion discontinuities and in textureless regions. In October 2007, we published the performance of several well-known methods on a preliminary version of our data to establish the current state of the art. We also made the data freely available on the web at . Subsequently a number of researchers have uploaded their results to our website and published papers using the data. A significant improvement in performance has already been achieved. In this paper we analyze the results obtained to date and draw a large number of conclusions from them. [ABSTRACT FROM AUTHOR]

Motion estimation is usually based on the brightness constancy assumption. This assumption holds well for rigid objects with a Lambertian surface, but it is less appropriate for fluid and gaseous materials. For these materials an alternative assumption is required. This work examines three possible alternatives: gradient constancy, color constancy and brightness conservation (under this assumption the brightness of an object can diffuse to its neighborhood). Brightness conservation and color constancy are found to be adequate models. We propose a method for detecting regions of dynamic texture in image sequences. Accurate segmentation into regions of static and dynamic texture is achieved using a level set scheme. The level set function separates each image into regions that obey brightness constancy and regions that obey the alternative assumption. We show that the method can be simplified to obtain a less robust but fast algorithm, capable of real-time performance. Experimental results demonstrate accurate segmentation by the full level set scheme, as well as by the simplified method. The experiments included challenging image sequences, in which color or geometry cues by themselves would be insufficient. [ABSTRACT FROM AUTHOR]

Inferring scene geometry from a sequence of camera images is one of the central problems in computer vision. While the overwhelming majority of related research focuses on diffuse surface models, there are cases when this is not a viable assumption: in many industrial applications, one has to deal with metal or coated surfaces exhibiting a strong specular behavior. We propose a novel and generalized constrained gradient descent method to determine the shape of a purely specular object from the reflection of a calibrated scene and additional data required to find a unique solution. This data is exemplarily provided by optical flow measurements obtained by small scale motion of the specular object, with camera and scene remaining stationary. We present a non-approximative general forward model to predict the optical flow of specular surfaces, covering rigid body motion as well as elastic deformation, and allowing for a characterization of problematic points. We demonstrate the applicability of our method by numerical experiments on synthetic and real data. [ABSTRACT FROM AUTHOR]

This paper describes a new approach to motion estimation in video. We represent video motion using a set of particles. Each particle is an image point sample with a long-duration trajectory and other properties. To optimize particle trajectories we measure appearance consistency along the particle trajectories and distortion between the particles. The resulting motion representation is useful for a variety of applications and cannot be directly obtained using existing methods such as optical flow or feature tracking. We demonstrate the algorithm on challenging real-world videos that include complex scene geometry, multiple types of occlusion, regions with low texture, and non-rigid deformations. [ABSTRACT FROM AUTHOR]

A novel optical flow estimation process based on a spatio-temporal model with varying coefficients multiplying a set of basis functions at each pixel is introduced. Previous optical flow estimation methodologies did not use such an over parameterized representation of the flow field as the problem is ill-posed even without introducing any additional parameters: Neighborhood based methods of the Lucas–Kanade type determine the flow at each pixel by constraining the flow to be described by a few parameters in small neighborhoods. Modern variational methods represent the optic flow directly via the flow field components at each pixel. The benefit of over-parametrization becomes evident in the smoothness term, which instead of directly penalizing for changes in the optic flow, accumulates a cost of deviating from the assumed optic flow model. Our proposed method is very general and the classical variational optical flow techniques are special cases of it, when used in conjunction with constant basis functions. Experimental results with the novel flow estimation process yield significant improvements with respect to the best results published so far. [ABSTRACT FROM AUTHOR]

In this paper we propose a new motion estimator for image sequences depicting fluid flows. The proposed estimator is based on the Helmholtz decomposition of vector fields. This decomposition consists in representing the velocity field as a sum of a divergence free component and a vorticity free component. The objective is to provide a low-dimensional parametric representation of optical flows by depicting them as deformations generated by a reduced number of vortex and source particles. Both components are approximated using a discretization of the vorticity and divergence maps through regularized Dirac measures. The resulting so called irrotational and solenoidal fields consist of linear combinations of basis functions obtained through a convolution product of the Green kernel gradient and the vorticity map or the divergence map respectively. The coefficient values and the basis function parameters are obtained by minimization of a functional relying on an integrated version of mass conservation principle of fluid mechanics. Results are provided on synthetic examples and real world sequences. [ABSTRACT FROM AUTHOR]

Traditional techniques of dense optical flow estimation do not generally yield symmetrical solutions: the results will differ if they are applied between images I 1 and I 2 or between images I 2 and I 1. In this work, we present a method to recover a dense optical flow field map from two images, while explicitely taking into account the symmetry across the images as well as possible occlusions in the flow field. The idea is to consider both displacements vectors from I 1 to I 2 and I 2 to I 1 and to minimise an energy functional that explicitely encodes all those properties. This variational problem is then solved using the gradient flow defined by the Euler–Lagrange equations associated to the energy. To prove the importance of the concepts of symmetry and occlusions for optical flow computation, we have extended a classical approach to handle those. Experiments clearly show the added value of these properties to improve the accuracy of the computed flows. Figures appear in color in the online version of this paper. [ABSTRACT FROM AUTHOR]

Autonomous mobile robot navigation, either off-road or on ill-structured roads, presents unique challenges for machine perception. A successful terrain or roadway classifier must be able to learn in a self-supervised manner and adapt to inter- and intra-run changes in the local environment. This paper demonstrates the improvements achieved by augmenting an existing self-supervised image segmentation procedure with an additional supervisory input. Obstacles and roads may differ in appearance at distance because of illumination and texture frequency properties. Reverse optical flow is added as an input to the image segmentation technique to find examples of a region of interest at previous times in the past. This provides representations of this region at multiple scales and allows the robot to better determine where more examples of this class appear in the image. [ABSTRACT FROM AUTHOR]

We present an analysis of the spatial and temporal statistics of “natural” optical flow fields and a novel flow algorithm that exploits their spatial statistics. Training flow fields are constructed using range images of natural scenes and 3D camera motions recovered from hand-held and car-mounted video sequences. A detailed analysis of optical flow statistics in natural scenes is presented and machine learning methods are developed to learn a Markov random field model of optical flow. The prior probability of a flow field is formulated as a Field-of-Experts model that captures the spatial statistics in overlapping patches and is trained using contrastive divergence. This new optical flow prior is compared with previous robust priors and is incorporated into a recent, accurate algorithm for dense optical flow computation. Experiments with natural and synthetic sequences illustrate how the learned optical flow prior quantitatively improves flow accuracy and how it captures the rich spatial structure found in natural scene motion. [ABSTRACT FROM AUTHOR]

Motris, an integrated system for model-based tracking research, has been designed modularly to study the effects of algorithmic variations on tracking results. Motris attempts to avoid introducing bias into the relative assessment of alternative approaches. Such a bias may be caused by differences of implementation and parameterization if the component approaches are evaluated in separate testing environments. Tracking results are evaluated automatically on a significant test sample in order to quantify the effects of different combinations of alternatives. The Motris system environment thus allows an in-depth comparison between the so-called ‘Edge-Element Association’ approach documented in Haag and Nagel (1999) and the more recent ‘Expectation-Maximization’ approach reported by Pece and Worrall (2002). [ABSTRACT FROM AUTHOR]

This special edition of the International Journal of Computer Vision is devoted to a selection of the best papers submitted to the 2018 British Machine Vision Conference (BMVC) held at Northumbria University, Newcastle upon Tyne, UK. Papers were assigned to Area Chairs (AC) and reviewers based on both AC and reviewer biddings on papers, CMT subject areas, and the Toronto Paper Matching System (TPMS) publication history. In " I Text2Sign: Towards Sign Language Production using Neural Machine Translation and Generative Adversarial Networks i ", Stoll et al. presented a system to automatically produce sign language videos from spoken language sentences. [Extracted from the article]

Detecting and estimating motions of fast moving objects has many important applications. However, most existing motion estimation techniques have difficulties in handling large motions in the scene. In this paper, we extend our recently proposed reliability-based stereo vision technique to solving large motion estimation problem. Compared with our stereo vision approach, the new algorithm removes the constant penalty assumption and explicitly enforces the inter-scanline consistency constraint. The resulting algorithm can handle sequences that contain large motions and can produce optical flows with 100% density over the entire image domain. The experimental results indicate that it can generate more accurate optical flows than existing approaches. [ABSTRACT FROM AUTHOR]