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

From Computer Vision Freiburg:  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.

his video shows impressions from various parts of our dataset, as well as state-of-the-art realtime disparity estimation results produced by one of our new CNNs... (full paper)

 

 

Comments (0)

This post does not have any comments. Be the first to leave a comment below.


Post A Comment

You must be logged in before you can post a comment. Login now.

Featured Product

Denso Robotics - Newest 6-axis VMB Series offers longer arm reach and higher load capacity

Denso Robotics - Newest 6-axis VMB Series offers longer arm reach and higher load capacity

The new VMB series represent some of the newest members to our 6-axis family of robots. These high-performance, versatile units offer a longer arm reach and a higher load capacity than traditional models which make VMB an excellent solution for palletizing, packaging, and material handling. New features include greater air piping, valve and signal line options, as well as new programming options with state-of-the-art functions using our new WINCAPS Plus software. VMB offers an IP67 protection grade along with meeting ISO Class 5, which makes them suitable for electric parts, food manufacturing processes, and pharmaceutical and medical devices. With the addition of the new VMB large robots, all manufacturing processes can now be automated by DENSO Robotics.