Target detection method and apparatus, and neural network training method and apparatus

Abstract

The present invention provides a target detection method and apparatus based on a neural network, and a neural network training method and apparatus for target detection. The target detection method comprises: acquiring a target-containing image to be detected; acquiring the first feature information of the image to be detected by employing a first pre-trained neural network, and acquiring the second feature information of the image to be detected by employing a second pre-trained neural network; combining the first feature information and the second feature information to acquire combined feature information; and acquiring a target detection result based on the combined feature information by employing the second neural network, wherein the number of layers of the second neural network is larger than the number of layers of the first neural network, and the first feature information is heat map feature information, and the second feature information is picture feature information.

Description

technical field [0001] The present disclosure relates to the field of image processing, and more specifically, the present disclosure relates to a neural network-based target detection method and device, and a neural network training method and device for target detection. Background technique [0002] Object detection is a basic research topic in the field of computer vision, and it has broad application prospects in many aspects such as face recognition, security monitoring, and dynamic tracking. Target detection refers to detecting and identifying a specific target (such as a pedestrian) in any given image, and returning the position and size information of the target, such as outputting a bounding box surrounding the target. Object detection is a complex and challenging pattern detection problem. On the one hand, due to internal changes such as target detail changes and occlusions, on the other hand, due to imaging angles, lighting effects, focal lengths of imaging devic...

AI Technical Summary

Problems solved by technology

There are two main problems that plague the current pedestrian target detection methods: one is that a large number of "false positive" detection results will be generated, that is, non-target areas are marked as targets; the other is that due to the influence of illumination and target posture, some targets cannot be detected. The neural network automatically detects

This is because the current training and detection process of the neural network used for target detection often directly generates the position of the target in the picture, and does not fully consider the splitting of this process and the iterative training of the network, nor does it consider Other factors that can assist training and improve detection accuracy

Images