Abstract: In view of the problem that the transmission line spans a wide area and has many hidden targets in the transmission channel, the visual grading warning model of the transmission line channel is put forward. First of all, the depth residual network to is improved to extract the multispectral information of the input image, the noise effect is reduced by soft threshold, the accuracy of the transmission line channel scene analysis model is improved, and then YOLOv3 target detection algorithm is used to build the transmission line channel hidden danger target identification model. Aiming at the problems of hidden smoke and small targets of contruction vihicles, hard example mining strategy is adopted to reduce the impact of the picture background, and then an early warning model is build based on the transmission line channel classification early warning structure. The results show that the visual and graded early warning model of transmission line channel combined with scene analysis can scientifically and accurately reflect the hidden danger warning state of transmission line channel and provide guidance for the operation and maintenance of transmission line.