Method and device for detecting performance of crowd analysis algorithm, equipment and storage medium
By automatically setting the recognition range and reference crowd height, and using crowd analysis algorithm models to identify abnormal events, this technology solves the problems of low efficiency and low accuracy in the detection of crowd analysis algorithms in the prior art, and achieves efficient and accurate performance evaluation.
CN116824311BActive Publication Date: 2026-07-07PING AN BANK CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- PING AN BANK CO LTD
- Filing Date
- 2023-07-04
- Publication Date
- 2026-07-07
AI Technical Summary
Technical Problem
Existing crowd analysis algorithms suffer from low performance, low accuracy, and high manpower and material costs.
Method used
By automatically extracting video frame images from the target video, setting the reference crowd height and recognition range, using a crowd analysis algorithm model to identify abnormal events, and recording the details of abnormal events to evaluate the model performance.
Benefits of technology
It achieves automated and accurate performance testing, reduces the investment of manpower and material resources, supports comprehensive testing in multiple scenarios, and improves testing efficiency and accuracy.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116824311B_ABST
Abstract
The application relates to artificial intelligence and discloses a performance detection method and device of a crowd analysis algorithm, equipment and a storage medium, which comprises the following steps: extracting video frame images from a target video; setting reference crowd height and an identification area of a target video frame image in the video frame images according to crowd height data and identification range data; inputting the extracted video frame images into a crowd analysis algorithm model to be detected, identifying abnormal events in the identification area in the target video by using the crowd analysis algorithm model to be detected, and outputting abnormal event alarm information when it is determined that there is an abnormal event; counting each kind of abnormal event according to the abnormal event alarm information to obtain a statistical result; and obtaining a performance detection result of the crowd analysis algorithm model to be detected according to the statistical result. The application can effectively determine the accuracy of the crowd analysis algorithm model, reduces the investment of manpower and material resources, and supports comprehensive testing of the model in various scenes.
Need to check novelty before this filing date? Find Prior Art