A Rainy Day Object Detection Method Based on YOLOv11 Combined with FFT and MF Model Fusion
Jin Haek Yang, Tao Tian, Yuanxun Liu, Chengshan Li, Dingping Wu, Lin Wang, Xiancheng Wang · 2024
At the entrance of the GCB room in a power station, flood protection grooves are installed. In case of emergencies, such as Rainstorm weather, manual installation of blocking boards, flood protection sandbags, and other materials is required outside the GCB room. This manual process is time-consuming, inefficient, difficult to carry out, and involves high labor risks. It may also lead to significant water leakage and poor sealing performance, potentially allowing water ingress that threatens the safe operation of equipment. Therefore, the development of an automatic flood protection sealing device is being explored. This device should be capable of remote and on-site deployment during emergencies, allowing rapid sealing to prevent rainwater from flooding the plant through the GCB room door. During Rainstorm flooding, the flood protection sealing device will be deployed outside the GCB room. When the flood barrier is raised during a Rainstorm flooding event, there may be objects or personnel on the barrier. To prevent hazards and avoid objects sliding into the hinge area, which could block its closure, this paper proposes a method to detect the presence of workers or equipment in designated areas under low-light conditions. The method integrates YOLOv11, FFT, MF, and model fusion to ensure reliable detection during rainy conditions.