Image detection method, device, computer equipment and storage medium

Abstract

The present invention proposes an image detection method, device, computer equipment, and storage medium, wherein the method includes: regularly segmenting the collected image to form a plurality of first image slices; inputting the first image slice into the first The target detection is carried out in the layer joint neural network to obtain the first detection result; wherein, the first detection result includes the first position information and the first category information of the first image slice; intercepting the image matching the first position information from the image The second image fragmentation; the second image fragmentation is input into the second layer joint neural network for target detection, and the second detection result is obtained; wherein, the second detection result includes the second position information and the second image fragmentation of the second image fragmentation Second category information: according to the second category information, determine the target category and the confidence level of the target category in the second image segment. Through the method, the number of areas to be detected is greatly reduced, the time-consuming detection can be shortened, and the detection efficiency and detection accuracy can be improved.

Description

technical field [0001] The invention relates to the technical field of rail transit, in particular to an image detection method, device, computer equipment and storage medium. Background technique [0002] In the field of rail transit, as the research enthusiasm for unmanned driving technology continues to rise, how to detect obstacles in front of the train to ensure the safe operation of the train has gradually become a research hotspot. [0003] At present, obstacles in front of the train are mainly identified through object detection. In related technologies, the Regions with Convolutional Neural Network Features (RCNN) algorithm is mainly used to realize target detection. [0004] However, the RCNN algorithm mainly detects larger targets, and the targets that need to be detected during the train running are relatively small. The detection effect of the RCNN algorithm is poor and the accuracy is low. In addition, due to the high speed of the train, the real-time require...

AI Technical Summary

Problems solved by technology

[0004] However, the RCNN algorithm mainly detects larger targets, and the targets that need to be detected during the train running are relatively small. The detection effect of the RCNN algorithm is poor and the accuracy is low.

In addition, due to the high speed of the train, the real-time requirements are high, and the RCNN algorithm has to select 2000-3000 candidate regions for each image for detection, which takes a long time and has low detection efficiency, making it difficult to meet the real-time requirements. Require

Images