Target identification method and device

Abstract

The invention relates to a target recognition method and device, and relates to the technical field of image recognition, and the method comprises the following steps: obtaining a first image shot at a first zoom ratio, determining to-be-recognized targets at a first recall ratio, and obtaining the first target information of each to-be-recognized target; acquiring a second image of the to-be-recognized target shot at a second scaling, so as to check the optimal threshold value to recognize the second image, and obtaining second target information; acquiring a third image of the to-be-identified target shot at a third scaling, and identifying the third image by checking the optimal threshold to obtain third target information; and obtaining final identification information according to the first target information, the second target information and the third target information. According to the method and device, the multiple of target detection is adjusted, different recall ratios and precision ratios are combined in the acquired images with different multiple, and the neural network model is utilized to perform image analysis, so that the precision ratio is ensured on the premise of ensuring the recall ratio, and the accuracy of target detection is improved.

Description

technical field [0001] The present application relates to the technical field of image recognition, in particular to a target recognition method and device. Background technique [0002] With the development of image recognition technology, the target detection technology in the industry has generally adopted the target detection method based on artificial intelligence. It can be known from practice that the target detection method based on artificial intelligence needs to balance the recall rate and the precision rate. The reason is that Input an image into the artificial intelligence neural network, and the target detection result given is the coordinates of the target in the image, the type of the target, and the corresponding credibility. [0003] Usually, a threshold is set to distinguish between right and wrong, but in fact, there will be a problem that the target detection conclusion is yes but the credibility is low, and the target detection conclusion is false but t...

AI Technical Summary

Problems solved by technology

[0003] Usually, a threshold is set to distinguish between right and wrong, but in fact, there will be a problem that the target detection conclusion is yes but the credibility is low, and the target detection conclusion is false but the credibility is slightly higher, so there must be problems The problem of balancing recall and precision

Images