Evaluation method, device, computer equipment and storage medium for target detection algorithm

Abstract

The present application relates to an evaluation method, device, computer equipment and storage medium of a target detection algorithm. The method includes: labeling all n objects in the image to be detected to obtain n label frames, wherein n is a positive integer, performing target detection on the image to be detected to obtain an initial detection result, wherein, if the target detection is performed on the image to be detected If the target is detected, then the initial detection result of includes the detection frame, and the missing image obtained by missing some or all of the n objects in the image to be detected is subjected to target detection, and the missing detection result is obtained, wherein, if the missing image When a target is detected during target detection, the missing detection result contains a detection frame, and the target detection algorithm is evaluated according to the initial detection result and / or the missing detection result, so that the target detection algorithm can be accurately evaluated.

Description

Technical field [0001] The present application relates to the field of target detection, and in particular, a method, apparatus, computer device, and storage medium involving a target detection algorithm. Background technique [0002] Target detection is the technology that detects the target from the image. The accuracy of the target detection algorithm directly affects the target detection effect. Therefore, how to intelligently accurately evaluate the quality of a target detection algorithm has always been a technician in the art. The focus of the study, and the performance evaluation of the target detection algorithm is an evaluation of the detection effect of the target, and therefore, the target detection algorithm can be evaluated by evaluating the detection effect. [0003] In the traditional method, when the accuracy of multi-objective detection is performed, it is necessary to first mark the number of targets in the image, and then, the target detection algorithm is use...

AI Technical Summary

Problems solved by technology

Although this method is fast, there are the following problems: First, although it is possible to judge whether there is a target missed detection or a non-target false detection by judging the quantitative relationship between the detected frame and the marked frame (when the number of detected frames is greater than the number of marked frames, non-target Target misdetection, when the number of detected frames is less than the number of labeled frames, target missed detection occurs), but it is impossible to determine which target is missed or which non-target is falsely detected; second, the detection frame is not always the same as The label box with the largest overlapping area corresponds, and it is not rigorous to use the overlapping area as the corresponding label

Images