DeepPruner: Learning Efficient Stereo Matching via Differentiable PatchMatch

TL;DR

DeepPruner introduces a differentiable PatchMatch approach that prunes disparity search space efficiently, enabling real-time stereo matching with competitive accuracy on KITTI and SceneFlow datasets.

Contribution

It presents a novel differentiable PatchMatch module that reduces computation by pruning disparities, allowing end-to-end training for real-time stereo matching.

Findings

Achieves real-time inference at 62ms per frame.

Maintains competitive accuracy on KITTI and SceneFlow datasets.

Reduces memory and computation through effective disparity pruning.

Abstract

Our goal is to significantly speed up the runtime of current state-of-the-art stereo algorithms to enable real-time inference. Towards this goal, we developed a differentiable PatchMatch module that allows us to discard most disparities without requiring full cost volume evaluation. We then exploit this representation to learn which range to prune for each pixel. By progressively reducing the search space and effectively propagating such information, we are able to efficiently compute the cost volume for high likelihood hypotheses and achieve savings in both memory and computation. Finally, an image guided refinement module is exploited to further improve the performance. Since all our components are differentiable, the full network can be trained end-to-end. Our experiments show that our method achieves competitive results on KITTI and SceneFlow datasets while running in real-time at 62ms.