In this paper, in order to obtain a better classification evaluation effect, a feedback connection model is added to the convolutional neural network to establish the evaluation model of the integration of industry and education in colleges and universities based on FCNN. Compare the MSE loss values of the traditional BP neural network model and the FCNN model. Indicator system construction, with the help of the accuracy of the convolutional neural network, to carry out the whole process of evaluation around the indicators, weights, and the quality of the implementation results. The data of students’ micro-expression concentration recognition test is used as the evaluation data of students’ project participation, comparing the recognition rate of the participation evaluation system proposed in this paper and the traditional participation evaluation system to complete the quality evaluation of the talent cultivation model of college education. Analyze the data on the graduation rates of college graduates to determine the effectiveness of the university’s integration of college education. The analysis shows that in 2022, the professional matching employment rate of graduates was 86.28%, which reflects the high efficiency of the university’s industry-teaching integration on the cultivation of professional and applied talents, and the mechanism of industry-teaching integration is well affiliated.