Integer-Valued Training and Spike-Driven Inference Spiking Neural Network for High-performance and Energy-efficient Object Detection

TL;DR

This paper introduces SpikeYOLO, a novel spiking neural network architecture with integer-valued training and spike-driven inference, achieving high-performance and energy-efficient object detection surpassing prior SNNs and comparable to ANNs.

Contribution

The paper proposes a simplified SpikeYOLO architecture and a new spiking neuron design that maintains integer values during training and inference, significantly improving object detection performance and energy efficiency.

Findings

Achieved 66.2% mAP@50 on COCO, +15.0% over previous SNNs.

Attained 67.2% mAP@50 on neuromorphic Gen1, +2.5% over ANN.

Energy efficiency improved by 5.7 times.

Abstract

Brain-inspired Spiking Neural Networks (SNNs) have bio-plausibility and low-power advantages over Artificial Neural Networks (ANNs). Applications of SNNs are currently limited to simple classification tasks because of their poor performance. In this work, we focus on bridging the performance gap between ANNs and SNNs on object detection. Our design revolves around network architecture and spiking neuron. First, the overly complex module design causes spike degradation when the YOLO series is converted to the corresponding spiking version. We design a SpikeYOLO architecture to solve this problem by simplifying the vanilla YOLO and incorporating meta SNN blocks. Second, object detection is more sensitive to quantization errors in the conversion of membrane potentials into binary spikes by spiking neurons. To address this challenge, we design a new spiking neuron that activates Integer values during training while maintaining spike-driven by extending virtual timesteps during inference. The proposed method is validated on both static and neuromorphic object detection datasets. On the static COCO dataset, we obtain 66.2% mAP@50 and 48.9% mAP@50:95, which is +15.0% and +18.7% higher than the prior state-of-the-art SNN, respectively. On the neuromorphic Gen1 dataset, we achieve 67.2% mAP@50, which is +2.5% greater than the ANN with equivalent architecture, and the energy efficiency is improved by 5.7*. Code: https://github.com/BICLab/SpikeYOLO