Aiming at the problem of invisible diseases of tower foundation concrete in power transmission and transformation projects caused by improper operation and related environmental factors during construction, this paper proposes a GPR image recognition method for hidden diseases of tower foundation in power transmission and transformation projects based on improved Faster R-CNN. Firstly, the traditional Faster R-CNN model is optimized, and RseNet-50 is used as the feature extraction backbone network, combined with the attention mechanism (SE module). By adding the SE module to different levels of the network and placing the SE module in different positions for comparative experiments, the optimal position in the model is judged, so that the model can reduce the redundant calculation burden on the basis of extracting the key information in the feature map. Secondly, the soft-NMS algorithm is used to improve the detection ability of closely connected targets. Finally, based on the measured data of the rigid straight column foundation, the simulation is carried out by gprMax, and the GPR image is processed by the generative adversarial network to expand the data set and enhance the learning ability of the model. The experimental results show that the average accuracy of the optimized model is 84.49%, and the F-Score is 77.58%. Compared with the traditional Faster-RCNN target detection model, the recognition accuracy is improved by 6.37% on the basis of the same data set.