Unmanned aerial vehicle ground detection system and method based on cascaded sparse query
The UAV surface detection system based on cascaded sparse queries solves the problem of low accuracy in small target detection in UAV surface detection by utilizing the FFM receptive field enhancement module, the AFF adaptive feature fusion module, and the improved Focal Loss loss function, thus achieving efficient and accurate multi-scale target detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- INSPUR QILU SOFTWARE IND
- Filing Date
- 2023-08-24
- Publication Date
- 2026-06-09
AI Technical Summary
In current UAV surface detection, the detection of small targets is not very effective, mainly due to problems such as the disappearance of small target features, mismatch of receptive fields, and difficulty in localization, resulting in low detection accuracy.
The UAV surface detection system employing cascaded sparse queries includes an FFM receptive field enhancement module, an AFF adaptive feature fusion module, a cascaded sparse query detection head module, and a class-independent positive and negative sample balancing strategy. It improves detection accuracy and speed through dilated convolution, sparse queries, and an improved Focal Loss loss function.
It significantly improves the accuracy of small target detection, reduces computational overhead, enhances the robustness and inference speed of the model, and is suitable for deployment on domestically produced AI accelerator cards.
Smart Images
Figure CN117132909B_ABST