ThunderNet: Towards Real-time Generic Object Detection

TL;DR

ThunderNet is a lightweight, efficient two-stage object detector designed for real-time performance on mobile devices, achieving high accuracy with low computational cost and running at 24.1 fps on ARM platforms.

Contribution

The paper introduces ThunderNet, a novel lightweight two-stage detector with a specialized backbone and detection modules optimized for real-time mobile detection.

Findings

Achieves superior performance with only 40% of the computational cost compared to other lightweight detectors.

Runs at 24.1 fps on ARM-based devices, demonstrating real-time capability.

First reported real-time detector on ARM platforms.

Abstract

Real-time generic object detection on mobile platforms is a crucial but challenging computer vision task. However, previous CNN-based detectors suffer from enormous computational cost, which hinders them from real-time inference in computation-constrained scenarios. In this paper, we investigate the effectiveness of two-stage detectors in real-time generic detection and propose a lightweight two-stage detector named ThunderNet. In the backbone part, we analyze the drawbacks in previous lightweight backbones and present a lightweight backbone designed for object detection. In the detection part, we exploit an extremely efficient RPN and detection head design. To generate more discriminative feature representation, we design two efficient architecture blocks, Context Enhancement Module and Spatial Attention Module. At last, we investigate the balance between the input resolution, the backbone, and the detection head. Compared with lightweight one-stage detectors, ThunderNet achieves superior performance with only 40% of the computational cost on PASCAL VOC and COCO benchmarks. Without bells and whistles, our model runs at 24.1 fps on an ARM-based device. To the best of our knowledge, this is the first real-time detector reported on ARM platforms. Our code and models are available at \url{https://github.com/qinzheng93/ThunderNet}.