In the recent years, due to the rapid growth in the number of unmanned aerial vehicles UAV and the continuous increase in task complexity, UAV networks have encountered severe challenges in data transmission latency, network congestion, signal non-uniformity, and other aspects. This article intends to conduct research on UAV network optimization and management based on machine vision. Firstly, real-time perception and recognition of the environment by UAV can be achieved through computer vision technology. Secondly, machine learning algorithms are combined to optimize the flight path and network resource allocation of UAV, and the communication frequency and power between UAV are dynamically adjusted through intelligent scheduling algorithms to reduce signal interference and improve network transmission efficiency. In experiments numbered 5, 7, and 15, the data transmission rates are increased from 30 Mbps, 35 Mbps, and 27 Mbps to 40 Mbps, 45 Mbps, and 37 Mbps, respectively. This article providing theoretical basis and practical guidance for technological development and innovation in related fields.