A rotating target detection method, device, equipment and storage medium

By employing feature enhancement and rotation angle fusion methods, the problem of missed detections and false detections in target detection networks with a small number of training samples was solved, achieving higher detection accuracy and precision.

CN116246056BActive Publication Date: 2026-07-07CHINA AUTOMOTIVE INNOVATION CORP
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
CHINA AUTOMOTIVE INNOVATION CORP
Filing Date
2022-09-07
Publication Date
2026-07-07

AI Technical Summary

Technical Problem

Existing object detection networks are prone to missed detections and false detections for object categories with few training samples (such as buses and fire trucks), resulting in insufficient detection accuracy.

Method used

By acquiring image feature information and initial detection box information, feature enhancement processing is performed using the first feature distribution information, and fusion processing is performed by combining the detection box rotation angle information. The result is then input into the rotating target detection network for detection, thereby improving the accuracy of the detection box.

Benefits of technology

It improves the accuracy of target category recognition and the precision of target detection boxes, reduces background intrusion, and enhances detection performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116246056B_ABST
    Figure CN116246056B_ABST
Patent Text Reader

Abstract

The application discloses a rotating target detection method and device, equipment and a storage medium. The method comprises the following steps: acquiring image feature information corresponding to a to-be-detected image, initial detection frame information and first category information of a detection target in the to-be-detected image, and the initial detection frame information comprises detection frame rotation angle information; performing feature enhancement processing on the image feature information based on first feature distribution information corresponding to the first category information, to obtain enhanced image feature information; performing fusion processing on the enhanced image feature information and the detection frame rotation angle information to obtain target image feature information; inputting the target image feature information into a rotating target detection network to perform rotating target detection, to obtain target detection frame information and second category information corresponding to the detection target, and the second category information is subcategory information under the first category information. The technical scheme of the application can improve the accuracy of detection target category identification and the precision of target detection frame information.
Need to check novelty before this filing date? Find Prior Art