Object detection plays an important role in self-driving cars for security development. However, mobile systems on self-driving cars with limited computation resources lead to difficulties for object detection. To facilitate this, we propose a compiler-aware neural pruning search framework to achieve high-speed inference on autonomous vehicles for 2D and 3D object detection. The framework automatically searches the pruning scheme and rate for each layer to find a best-suited pruning for optimizing detection accuracy and speed performance under compiler optimization. Our experiments demonstrate that for the first time, the proposed method achieves (close-to) real-time, 55ms and 99ms inference times for YOLOv4 based 2D object detection and PointPillars based 3D detection, respectively, on an off-the-shelf mobile phone with minor (or no) accuracy loss.
@article{arxiv.2106.14943,
title = {Achieving Real-Time Object Detection on MobileDevices with Neural Pruning Search},
author = {Pu Zhao and Wei Niu and Geng Yuan and Yuxuan Cai and Bin Ren and Yanzhi Wang and Xue Lin},
journal= {arXiv preprint arXiv:2106.14943},
year = {2021}
}
Comments
Presented on the HiPEAC 2021 workshop (cogarch 2021)