In this project, a video analytics framework for smart advertisement has been implemented. The proposed project has two major components. The video analytics component and the advert system component. The analytics framework includes an LSTM module which is used to perform a time-series prediction of the age group and gender that might be in a particular location when there is no face detection in a frame. The evaluation results of the module have shown that the model achieves a 99.99% accuracy on the validation dataset.
-
Retail dataset with customer transactions. including gender DOB and purchased item link