A lightweight context fusion object detection network
By embedding a region context enhancement module and a self-weighted fusion module into a lightweight object detection network, the problems of insufficient context awareness and single multi-scale feature fusion in embedded devices are solved, improving the robustness and detection accuracy of the model and making it suitable for embedded devices.
Patent Information
- Application Number
- CN202610838263.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-11
- Publication Date
- 2026-07-17
AI Technical Summary
Existing lightweight object detection models suffer from insufficient context awareness, limited multi-scale feature fusion strategies, and poor robustness to complex environments on embedded devices, making it difficult to meet the requirements of low power consumption, low latency, and small memory usage.
A lightweight context fusion object detection network is designed. By embedding a region context enhancement module in the backbone network and a self-weighted fusion module in the neck network, local details and global context representations are enhanced, and adaptive multi-scale feature fusion is achieved to improve the robustness of the model.
It effectively enhances the feature representation capability for small and occluded targets, achieves more accurate multi-scale feature fusion, improves the robustness of the model in noisy and occluded environments, and has extremely small parameter and computational requirements, making it suitable for embedded devices.
Smart Images
Figure CN122416205A_ABST