Aiming at the problems that birds affect the efficiency of photovoltaic power generation in mountainous areas and destroy photovoltaic power generation equipment, a bird repellent system combined with computer vision is designed. Firstly, the hardware design of the intelligent bird repellent system is introduced. The Jetson Nano microcomputer is used as the upper computer combined with the monitoring camera in the photovoltaic area to control the work of the bird repellent, which can achieve long-term bird repellent effect. Second, for the detection of birds, an improved yolov8 neural network algorithm based on attention mechanism is designed. The ContextAggregation attention module was introduced behind the C2f module in the neck of the YOLOv8 network to increase the sensitivity of the network to small targets. The improved network model can accurately predict the little bird target and the blocked target in the complex environment, and solve the problem of missing the small target and the blocked target in the original model. The experimental results show that the proposed method can effectively detect birds in the photovoltaic area of mountainous areas, provide data for the upper computer to control the bird repellent for bird repelling, and achieve long-term stable bird repelling effect.