Convolutional neural networks (CNNs) have recently been very successful in a variety of computer vision tasks, especially on those linked to recognition. Optical flow estimation has not been among the tasks CNNs succeeded at. In this paper we construct CNNs which are capable of solving the optical flow estimation problem as a supervised learning task. We propose and compare two architectures: a generic architecture and another one including a layer that correlates feature vectors at different image locations. Since existing ground truth data sets are not sufficiently large to train a CNN, we generate a large synthetic Flying Chairs dataset. We show that networks trained on this unrealistic data still generalize very well to existing datasets such as Sintel and KITTI, achieving competitive accuracy at frame rates of 5 to 10 fps.

/pdf/flownet-learning-optical-flow-with-convolutional-networks-2z6d44ur0r.pdf

FlowNet: Learning Optical Flow with Convolutional Networks

The FlowNet demonstrated that optical flow estimation can be cast as a learning problem. However, the state of the art with regard to the quality of the flow has still been defined by traditional methods. Particularly on small displacements and real-world data, FlowNet cannot compete with variational methods. In this paper, we advance the concept of end-to-end learning of optical flow and make it work really well. The large improvements in quality and speed are caused by three major contributions: first, we focus on the training data and show that the schedule of presenting data during training is very important. Second, we develop a stacked architecture that includes warping of the second image with intermediate optical flow. Third, we elaborate on small displacements by introducing a subnetwork specializing on small motions. FlowNet 2.0 is only marginally slower than the original FlowNet but decreases the estimation error by more than 50%. It performs on par with state-of-the-art methods, while running at interactive frame rates. Moreover, we present faster variants that allow optical flow computation at up to 140fps with accuracy matching the original FlowNet.

/pdf/flownet-2-0-evolution-of-optical-flow-estimation-with-deep-3xjp4o4y7c.pdf

FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networks

We present a compact but effective CNN model for optical flow, called PWC-Net. PWC-Net has been designed according to simple and well-established principles: pyramidal processing, warping, and the use of a cost volume. Cast in a learnable feature pyramid, PWC-Net uses the current optical flow estimate to warp the CNN features of the second image. It then uses the warped features and features of the first image to construct a cost volume, which is processed by a CNN to estimate the optical flow. PWC-Net is 17 times smaller in size and easier to train than the recent FlowNet2 model. Moreover, it outperforms all published optical flow methods on the MPI Sintel final pass and KITTI 2015 benchmarks, running at about 35 fps on Sintel resolution (1024 A— 436) images. Our models are available on our project website.

/pdf/pwc-net-cnns-for-optical-flow-using-pyramid-warping-and-cost-3opp05vuvh.pdf

PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume

Recent work has shown that optical flow estimation can be formulated as a supervised learning task and can be successfully solved with convolutional networks. Training of the so-called FlowNet was enabled by a large synthetically generated dataset. The present paper extends the concept of optical flow estimation via convolutional networks to disparity and scene flow estimation. To this end, we propose three synthetic stereo video datasets with sufficient realism, variation, and size to successfully train large networks. Our datasets are the first large-scale datasets to enable training and evaluating scene flow methods. Besides the datasets, we present a convolutional network for real-time disparity estimation that provides state-of-the-art results. By combining a flow and disparity estimation network and training it jointly, we demonstrate the first scene flow estimation with a convolutional network.

A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimation

Over the years, datasets and benchmarks have proven their fundamental importance in computer vision research, enabling targeted progress and objective comparisons in many fields. At the same time, legacy datasets may impend the evolution of a field due to saturated algorithm performance and the lack of contemporary, high quality data. In this work we present a new benchmark dataset and evaluation methodology for the area of video object segmentation. The dataset, named DAVIS (Densely Annotated VIdeo Segmentation), consists of fifty high quality, Full HD video sequences, spanning multiple occurrences of common video object segmentation challenges such as occlusions, motionblur and appearance changes. Each video is accompanied by densely annotated, pixel-accurate and per-frame ground truth segmentation. In addition, we provide a comprehensive analysis of several state-of-the-art segmentation approaches using three complementary metrics that measure the spatial extent of the segmentation, the accuracy of the silhouette contours and the temporal coherence. The results uncover strengths and weaknesses of current approaches, opening up promising directions for future works.

/pdf/a-benchmark-dataset-and-evaluation-methodology-for-video-4eb3ugu88k.pdf

A Benchmark Dataset and Evaluation Methodology for Video Object Segmentation

Ground truth optical flow is difficult to measure in real scenes with natural motion As a result, optical flow data sets are restricted in terms of size, complexity, and diversity, making optical flow algorithms difficult to train and test on realistic data We introduce a new optical flow data set derived from the open source 3D animated short film Sintel This data set has important features not present in the popular Middlebury flow evaluation: long sequences, large motions, specular reflections, motion blur, defocus blur, and atmospheric effects Because the graphics data that generated the movie is open source, we are able to render scenes under conditions of varying complexity to evaluate where existing flow algorithms fail We evaluate several recent optical flow algorithms and find that current highly-ranked methods on the Middlebury evaluation have difficulty with this more complex data set suggesting further research on optical flow estimation is needed To validate the use of synthetic data, we compare the image- and flow-statistics of Sintel to those of real films and videos and show that they are similar The data set, metrics, and evaluation website are publicly available

/pdf/a-naturalistic-open-source-movie-for-optical-flow-evaluation-xaa5syym2h.pdf

A naturalistic open source movie for optical flow evaluation

Methods, systems and media are described for computer-assisted video surveillance. Methods may support detection of moving persons in video frames, extraction of features of the detected moving persons and identification of which detected moving persons are likely matches to a person of interest. Identification of the likely matches may be determined using an attribute-based search, and/or using a specific person-based search. The method may include using likely matches confirmed as images of the person of interest to reconstruct a path of the person of interest.

Assisted video surveillance of persons-of-interest

Though teleoperated robots have become common for more extreme tasks such as bomb diffusion, search-and-rescue, and space exploration, they are not commonly used in human-populated environments for more ordinary tasks such as house cleaning or cooking. This presents near-term opportunities for teleoperated robots in the home. However, a teleoperator’s remote presence in a consumer’s home presents serious security and privacy risks, and the concerns of end-users about these risks may hinder the adoption of such in-home robots. In this paper, we define and explore the privacy-utility tradeoff for remotely teleoperated robots: as we reduce the quantity or fidelity of visual information received by the teleoperator to preserve the end-user’s privacy, we must balance this against the teleoperator’s need for sufficient information to successfully carry out tasks. We explore this tradeoff with two surveys that provide a framework for understanding the privacy attitudes of end-users, and with a user study that empirically examines the effect of different filters of visual information on the ability of a teleoperator to carry out a task. Our findings include that respondents do desire privacy protective measures from teleoperators, that respondents prefer certain visual filters from a privacy perspective, and that, for the studied task, we can identify a filter that balances privacy with utility. We make recommendations for in-home teleoperation based on these findings. Categories and Subject Descriptors H.1.2 [Models and Principles]: User/Machine Systems— human factors, software psychology General Terms Design; Human Factors

/pdf/the-privacy-utility-tradeoff-for-remotely-teleoperated-4agakem0iu.pdf

The Privacy-Utility Tradeoff for Remotely Teleoperated Robots

With the MPI-Sintel Flow dataset, we introduce a naturalistic dataset for optical flow evaluation derived from the open source CGI movie Sintel. In contrast to the well-known Middlebury dataset, the MPI-Sintel Flow dataset contains longer and more varied sequences with image degradations such as motion blur, defocus blur, and atmospheric effects. Animators use a variety of techniques that produce pleasing images but make the raw animation data inappropriate for computer vision applications if used "out of the box". Several changes to the rendering software and animation files were necessary in order to produce data for flow evaluation and similar changes are likely for future efforts to construct a scientific dataset from an animated film. Here we distill our experience with Sintel into a set of best practices for using computer animation to generate scientific data for vision research.

/pdf/lessons-and-insights-from-creating-a-synthetic-optical-flow-wi9i0bjni1.pdf

Lessons and insights from creating a synthetic optical flow benchmark

We introduce an approach for analyzing Wikipedia and other text, together with online photos, to produce annotated 3D models of famous tourist sites. The approach is completely automated, and leverages online text and photo co-occurrences via Google Image Search. It enables a number of new interactions, which we demonstrate in a new 3D visualization tool. Text can be selected to move the camera to the corresponding objects, 3D bounding boxes provide anchors back to the text describing them, and the overall narrative of the text provides a temporal guide for automatically flying through the scene to visualize the world as you read about it. We show compelling results on several major tourist sites.

/pdf/3d-wikipedia-using-online-text-to-automatically-label-and-2q4kgophrh.pdf

Daniel J. Butler

Papers

A naturalistic open source movie for optical flow evaluation

Assisted video surveillance of persons-of-interest

The Privacy-Utility Tradeoff for Remotely Teleoperated Robots

Lessons and insights from creating a synthetic optical flow benchmark

3D Wikipedia: using online text to automatically label and navigate reconstructed geometry