Abstract: Bird nesting on the transmission line towers will bring a great threat to the stable operation of power grids. Aimed at the problem of bird's nest detection in aerial images taken by UAV, a detection method for bird's nest on the transmission line towers based on visual saliency is proposed. The area of transmission tower was located by gray value analysis and is regarded as the region of interest. Three salient features were extracted from the region of interest in the image, and the conditional random field model was adopted to fuse these three salient features to obtain the saliency map. The constraint set by the shape of bird's nest was used to eliminate pseudo-targets such as power lines and clouds. Experimental results show that the proposed algorithm can effectively detect the bird's nest on the transmission line tower.