The invention discloses a face recognition method for advertising screens based on self-learning, which includes the step of secondary merging, learning the sample eigenvalues accumulated in the cycle time, and performing secondary screening on the eigenvalues belonging to the same face angle interval Check, merge the feature values of the same face ID; use the face recognition algorithm to judge the similarity of the face features, if the similarity is greater than the set value, merge the recognized face features under the same face ID. The present invention realizes accurate data statistics through the self-learning method, thereby providing more accurate audience analysis for placing advertisements on the display screen. The present invention improves the recognition rate through autonomous learning, and the present invention can also continuously improve the recall rate of recognition through the study of sample feature values, and more accurate face recognition can improve the advertiser's understanding of the audience, making personalized advertisement delivery more accurate , and at the same time provide more reliable data support for calculating the effect of advertising.