Fast Disparity Estimation Using Dense Networks

By Rowel O. Atienza. 2018

29 citations recorded by the source corpus.

Disparity estimation is a difficult problem in stereo vision because the correspondence technique fails in images with textureless and repetitive regions. Recent body of work using deep convolutional neural networks (CNN) overcomes this problem with semantics. Most CNN implementations use an autoencoder method; stereo images are encoded, merged and finally decoded to predict the disparity map. In this paper, we present a CNN implementation inspired by dense networks to reduce the number of parameters. Furthermore, our approach takes into account semantic reasoning in disparity estimation. Our proposed network, called DenseMapNet, is compact, fast and can be trained end-to-end. DenseMapNet requires 290k parameters only and runs at 30Hz or faster on color stereo images in full resolution. Experimental results show that DenseMapNet accuracy is comparable with other significantly bigger CNN-based methods.

Subfields: Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Kaiyong Zhao
2. Xiaowen Chu
3. Myron Z. Brown
4. Gregory Hager
5. Shi-gang Zheng
6. Kevin Foster
7. Marc Bosch
8. Sean Wang

In Media Technology

1. Gregory Hager
2. Shi-gang Zheng
3. Kaiyong Zhao
4. Fei Deng
5. Qiang Wang
6. Gordon A. Christie
7. Xiaowen Chu

Citations in the corpus

• IRS: A Large Synthetic Indoor Robotics Stereo Dataset for Disparity and Surface Normal Estimation (2019, methodology): “It provides 1,064 stereo images with high-quality disparity maps and has been used to train effective networks on realistic data in [15], [2], [1].”
• Semantic Stereo for Incidental Satellite Images (2018, background): “To demonstrate the value of semantic category as a prior for stereo correspondence, we adapted DenseMapNet to include four input channels with the fourth channel being a classification label from semantic segmentation.”