A novel fault segment location based on deep network with transfer learning for the distribution network is proposed. At first, the wavelet packet transform (WPT) is adopted to decompose electric signals. The wavelet packet coefficients of each node are rearranged from low frequency to high frequency to obtain the time-frequency matrix. The time-frequency matrix can be converted into the pixel matrix with the property of the image by the color-coding. The pixel matrix can contain the working conditions of the current system. Then, transfer learning is performed on the AlexNet model, and the network structure is adjusted to adapt to distribution network fault segment identification. The fine-tune AlexNet network can autonomously extract the pixel matrix features as predictive variables. Finally, the pattern recognition algorithms of GRU, LVQ, NBC, ELM, and SVM are used to classify the fault features, and the fault area location for the distribution network is completed. Experimental analysis is carried out for the overhead/cable hybrid line with multi-branches. The classifying effects of the five pattern recognition algorithms are compared. The accuracy of the GRU algorithm is 99.92%. The testing results show that the proposed method is not affected by fault time, fault type, grounding resistance, and other factors. It can meet the fault location accuracy and reliability requirement of the distribution network.