Hierarchical Neural Architecture Search for Deep Stereo Matching
*Xuelian Cheng1,5, *Yiran Zhong2,6, Mehrtash Harandi1,7, Yuchao Dai3, Xiaojun Chang1, Tom Drummond1, Hongdong Li2,6, Zongyuan Ge1,4,5 1Faculty of Engineering, Monash University, 2Australian National University, 3Northwestern Polytechnical University, 4eResearch Centre, Monash University, 5Airdoc Research Australia, 6ACRV, 7Data61, CSIRO {xuelian.cheng, {xiaojun.chang, tom.drummond, {yiran.zhong,
Abstract
To reduce the human efforts in neural network design, Neural Architecture Search (NAS) has been applied with remarkable success to various high-level vision tasks such as classification and semantic segmentation. The underlying idea for the NAS algorithm is straightforward, namely, to enable the network the ability to choose among a set of operations (e.g.,convolution with different filter sizes), one is able to find an optimal architecture that is better adapted to the problem at hand. However, so far the success of NAS has not been enjoyed by low-level geometric vision tasks such as stereo matching. This is partly due to the fact that state-of-the-art deep stereo matching networks, designed by humans, are already sheer in size. Directly applying the NAS to such massive structures is computationally prohibitive based on the currently available mainstream computing resources. In this paper, we propose the first end-to-end hierarchical NAS framework for deep stereo matching by incorporating task-specific human knowledge into the neural architecture search framework. Specifically, following the gold standard pipeline for deep stereo matching (ie.,, feature extraction – feature volume
中文速览
立体匹配（stereo matching）是根据双目图像估计稠密视差图的经典视觉任务，但现有最优深度网络体积庞大、内存消耗极高，导致神经架构搜索（Neural Architecture Search, NAS）难以直接用于该任务。为此，本文提出 LEAStereo——首个面向深度立体匹配的端到端层次化 NAS 框架，核心思路是将立体匹配的几何先验知识（特征提取→特征体构建→稠密匹配的流水线）嵌入搜索空间，同时在"细胞级"和"网络级"两个层次上联合搜索特征网络与匹配网络的最优结构，从而大幅压缩搜索所需的计算代价。大量实验表明，LEAStereo 在 KITTI 2012、KITTI 2015、Middlebury 及 SceneFlow 四个权威基准上均排名第一，同时网络参数量更小、推理速度更快，全面超越此前所有人工设计网络和已有 NAS 方法。这项工作证明，将领域知识与自动架构搜索相结合，可以突破低层几何视觉任务中应用 NAS 的计算瓶颈，为该方向的后续研究提供了重要参考。
原文 arXiv:2010.13501;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2010.13501v1