Abstract: Co-seismic response identification of fixed-point deformation data currently relies on manual picking, and no automatic co-seismic response detection method has yet been put into application. The paper propose the first deep learning model for co-seismic response detection of fixed-point deformation data in China, which is used to detect co-seismic response signals quickly and accurately on the second data of a single Vertical Pendulum broadband tiltmeter. The model is constructed using the transfer learning technique, introducing three representative pre-trained models for earthquake detection on seismic data as feature extractors, migrating their knowledge and capabilities of earthquake detection on seismic data to fixed-point deformation data, and then designing and adapting the supporting data converters and classifiers. Tests on real observational data show that the model has good detection performance. The application on the continuous data from Jixian station proves that the model is not only able to detect all the co-seismic response events recorded manually, but also able to find more events that were not recognized manually, with an accuracy rate of no less than 75%, and the detection efficiency, detection capability and consistency are greatly improved compared with the traditional manual processing.