Slim-neck by GSConv: A lightweight-design for real-time detector architectures

TL;DR

This paper introduces GSConv, a new lightweight convolutional technique, and Slim-Neck (SNs), a design based on GSConv, to improve real-time object detection accuracy and efficiency on edge devices.

Contribution

The paper presents GSConv, a novel lightweight convolution method, and a Slim-Neck design that together enhance real-time detection performance with high accuracy and computational efficiency.

Findings

GSConv achieves a good balance between accuracy and speed.

SNs significantly improve real-time detector performance.

State-of-the-art results of 70.9% AP50 at ~100FPS on Tesla T4.

Abstract

Real-time object detection is significant for industrial and research fields. On edge devices, a giant model is difficult to achieve the real-time detecting requirement and a lightweight model built from a large number of the depth-wise separable convolutional could not achieve the sufficient accuracy. We introduce a new lightweight convolutional technique, GSConv, to lighten the model but maintain the accuracy. The GSConv accomplishes an excellent trade-off between the accuracy and speed. Furthermore, we provide a design suggestion based on the GSConv, Slim-Neck (SNs), to achieve a higher computational cost-effectiveness of the real-time detectors. The effectiveness of the SNs was robustly demonstrated in over twenty sets comparative experiments. In particular, the real-time detectors of ameliorated by the SNs obtain the state-of-the-art (70.9% AP50 for the SODA10M at a speed of ~ 100FPS on a Tesla T4) compared with the baselines. Code is available at https://github.com/alanli1997/slim-neck-by-gsconv