可见光相机通信, LED 阵列, 目标检测, YOLOv11 算法

We addresses the current problems of complex network structure, large number of parameters, and high computational complexity of the target LED( Light Emitting Diode) array detection algorithm is studied based on deep learning in the OCC(Optical Camera Communication) system. A detection algorithm for LED arrays based on Effeps-YOLOv11 is proposed. In the backbone network of Effeps-YOLOv11 (Effeps-You Only Look Once version 11) feature extraction, a lightweight EfficientNetV2 network is adopted to balance the network width, depth, and image resolution. The original complex attention module is replaced with the ECA (Efficient Channel Attention ) attention mechanism to simplify the network structure. A lightweight C3PC ( C3 Part Convolution) module is designed to reduce the computational complexity. And the Shape_IoU loss function is used to improve the positioning accuracy of the bounding box and enhance the accuracy of LED array positioning,providing an early guarantee for correct decoding. Currently, no public dataset has been established in the field of target LED array in the OCC system. The experiments are based on the OCC system experimental platform to collect data and establish the required training dataset. The experimental results show that the Effeps-YOLOv11 algorithm proposed in this paper can meet the requirements of the target LED array detection task in complex outdoor environments.

optical camera communication, light emitting diode(LED) array, object detection, you only look once version 11(YOLOv11)