In the complex electromagnetic environmentthe recognition rate of radar jamming signal is low under the condition of low Jamming Noise Ratio(JNR)and it is difficult to obtain a large number of jamming samples for training.To solve the problema radar jamming perception method based on Attention mechanismtransfer learning and residual neural network is proposed.The model takes the time-frequency domain data of interference and echo as inputautomatically extracts features and makes type judgment.The experimental results show that the model realizes the effective perceptual recognition of radar jamming under the condition of low JNR and small-sample training.The Attention mechanism and transfer learning can effectively improve the accuracy of perceptual recognition.The recognition performance is more robust and accurate than that of the traditional machine learning model and the unmodified neural network model.