ADCPNet: Adaptive Disparity Candidates Prediction Network for Efficient Real-Time Stereo Matching

TL;DR

ADCPNet introduces a two-stage, adaptive disparity prediction framework with a dynamic offset module and disparity-independent convolution, achieving efficient real-time stereo matching suitable for mobile devices.

Contribution

The paper presents a novel two-stage stereo matching network with dynamic offset prediction and disparity-independent convolution, improving accuracy and efficiency over existing lightweight models.

Findings

Outperforms state-of-the-art lightweight models in accuracy and speed

Effective on multiple datasets and platforms, especially mobile devices

Reduces computation stages while maintaining high disparity correction quality

Abstract

Efficient real-time disparity estimation is critical for the application of stereo vision systems in various areas. Recently, stereo network based on coarse-to-fine method has largely relieved the memory constraints and speed limitations of large-scale network models. Nevertheless, all of the previous coarse-to-fine designs employ constant offsets and three or more stages to progressively refine the coarse disparity map, still resulting in unsatisfactory computation accuracy and inference time when deployed on mobile devices. This paper claims that the coarse matching errors can be corrected efficiently with fewer stages as long as more accurate disparity candidates can be provided. Therefore, we propose a dynamic offset prediction module to meet different correction requirements of diverse objects and design an efficient two-stage framework. Besides, we propose a disparity-independent convolution to further improve the performance since it is more consistent with the local statistical characteristics of the compact cost volume. The evaluation results on multiple datasets and platforms clearly demonstrate that, the proposed network outperforms the state-of-the-art lightweight models especially for mobile devices in terms of accuracy and speed. Code will be made available.