To address the issues of blurred scenes, high complexity, and significant differences in surgical processes in surgical videos, as well as the inefficiencies of traditional surgical process recognition methods that rely on manual annotations, this study proposes an automatic surgical process recognition method based on Convolutional Neural Networks (CNN) and Bi-directional Long Short-Term Memory (Bi-LSTM) networks. This method utilizes CNNs to extract spatial features from surgical videos and employs Bi-LSTM networks to capture temporal dependencies, solving the misrecognition problem caused by high correlation between adjacent frames. To enhance the model"s generalization ability, median filtering and data augmentation techniques are applied, and training efficiency is improved using the Adam optimizer. Validation on the Cholec80 dataset shows that the surgical process recognition accuracy reaches 96.16% in offline mode and 91.84% in online mode. The experimental results indicate that the proposed model achieves high accuracy in surgical process recognition and provides significant solutions for surgical robot process recognition.