Low Latency of Object Detection for Spiking Neural Network

Spiking Neural Networks (SNNs), as a third-generation neural network, are well-suited for edge AI applications due to their binary spike nature. However, when it comes to com-plex tasks like object detection, SNN s often require a substan-tial number of time steps to achieve high performance. This limitation significantly hampers the widespread adoption of SNN s in latency-sensitive edge devices. In this paper, our fo-cus is on generating highly accurate and low-latency SNNs specifically for object detection. Firstly, we systematically derive the conversion between SNNs and ANNs and analyze how to improve the consistency between them: improving the spike firing rate and reducing the quantization error. Then we propose a structural replacement, quantization of ANN acti-vation and residual fix to allevicate the disparity. We evalu-ate our method on challenging dataset MS COCO, PASCAL VOC and our spike dataset. The experimental results show that the proposed method achieves higher accuracy and lower latency compared to previous work Spiking-YOLO. The ad-vantages of SNNs processing of spike signals are also demon-strated.