Brain-like target tracking method based on spiking neural network

Abstract

The invention provides a brain-like target tracking method based on a spiking neural network. The target detection model constructed by the invention is composed of a basic network and different superimposed convolutional layers, wherein the basic network is a spiking neural network, and the method comprises the steps: carrying out feature extraction on an input image; obtaining a multi-scale detection feature map through different superimposed convolution layers, wherein the large-scale feature map is used for detecting a small object, and the small-scale feature map is used for detecting a large object; and finally outputting a target coordinate position and a confidence probability based on the target detection model, and realizing tracking of the target by the mobile terminal through PID control. The invention has the beneficial effects that compared with a traditional target detection model, the pulse neural network based on the LIF neurons is adopted as the basic network, so thatthe target detection model has better biological rationality.

Description

technical field [0001] The invention relates to the field of target tracking, in particular to a brain-like target tracking method based on a pulse neural network. Background technique [0002] In recent years, with the continuous development of artificial intelligence technology, target tracking technology has made further progress, and it is widely used in industry, medical care, home furnishing, transportation, etc. An effective target detection model is very important for target tracking technology. The target detection model mainly detects the target on the input image and outputs information such as the position coordinates of the detected target, and secondly realizes the tracking of the target by the mobile terminal through the motion control algorithm. [0003] At present, deep learning is widely used in the field of target tracking, among which traditional target detection models such as Faster R-CNN, SSD and YOLO are more common. Although these object detection m...

AI Technical Summary

Problems solved by technology

Although these object detection models can use deep neural networks to achieve object detection on input images, the deep neural networks do not have good biological rationality, and there are hardware acceleration platforms such as large calculations, high power consumption, and excessive reliance on GPUs. And other constraints, which lead to slow target detection speed and low accuracy

Images