算法评估方法、装置、设备及存储介质
By annotating the target objects in images in dense scenes as a whole or individually, and combining the overlap calculation, detection and matching results are generated, which solves the problem of inaccurate algorithm evaluation results in dense scenes and improves the accuracy of algorithm evaluation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHENGDU INTELLIFUSION TECH CO LTD
- Filing Date
- 2022-12-23
- Publication Date
- 2026-07-17
AI Technical Summary
In dense scenarios, existing algorithm evaluation methods suffer from inaccurate or incorrect statistical results due to overlapping and intersecting targets, which affects the accuracy of the algorithm evaluation results.
By labeling overlapping target objects in the image to be processed as a whole or labeling individual non-overlapping target objects, a group of labeled objects is established. A preset recognition algorithm is used to identify detection boxes, the overlap between the labeled boxes and the detection boxes is calculated, the detection matching result is determined, and finally the algorithm evaluation result is generated.
It improves the accuracy of algorithm evaluation results, reduces the difference between evaluation data and actual data, and enhances the accuracy of matching detection boxes and annotation boxes.
Smart Images
Figure CN116071639B_ABST