Object detection method based on multi-path dense feature fusion fully convolutional network

Abstract

The invention relates to a target detection method based on multi-path dense feature fusion full convolutional network, using deep convolutional neural network to extract layered multi-scale feature maps with different feature information; using bottom-up bypass connection to bottom-up Top-down feature fusion; top-down dense feature fusion using top-down dense bypass connections; constructing target candidate boxes of different sizes and aspect ratios; using binary classifiers to reduce simple backgrounds in target candidate boxes samples, and jointly optimize a binary classifier, a multi-class classifier, and a bounding box regressor using a multi-task loss function. The present invention extracts image features based on a deep convolutional neural network, uses a multi-path dense feature fusion method to improve feature expression capabilities, constructs a fully convolutional network for target detection, and proposes joint optimization for reducing redundant simple background samples and multi-task losses The strategy improves the detection accuracy of the algorithm and obtains good target detection results.

Description

technical field [0001] The invention belongs to the technical field of computer vision target detection, in particular to a target detection method based on multi-path dense feature fusion full convolution network. Background technique [0002] In the perception engineering of human beings in the material world, more than 80% of the information comes from vision. For humans, images and videos are vivid and realistic descriptions of objective things, and they are also important multimedia information carriers. As one of the core research topics in the field of computer vision, target detection technology extracts target features through analysis, and then obtains the category and location information of the target. Object detection technology integrates cutting-edge technologies in many fields such as image processing, pattern recognition, artificial intelligence, computer vision, etc. Wide range of applications. [0003] Target detection technology is to extract the featu...

AI Technical Summary

Problems solved by technology

[0005] To sum up, although the target detection algorithm has achieved good results after decades of development, and the emergence of convolutional neural networks has improved the target detection accuracy a lot, many problems still need to be improved, for example, how to improve Effectively enrich target feature information, how to reduce redundant simple background samples, etc.

Images