Automatic Detection of Brand Logos is a tool identifying brand logos in images and videos, and calculate how long the logo is visible. It can not only identify 10 different brands, 5 international brands and 5 Chinese brands, (Adidas, Kappa, New Balance, Nike, Puma, 361, Anta, Erke, Lining, Xtep) but also locate the logo that has not been seen before, the logo that has not been trained. Given the specific logo image, the system can mark the location of the logo in the pictures and videos. It can help evaluate marketing campaigns, capture user reviews of the product, counterfeit detection and protect brands' intellectual property, personalize product recommendations, improve search algorithms.
Tensorflow 1.15.0
Keras 2.1.5
Trained weight of this project: https://drive.google.com/file/d/13tMnidB3J1rdIpfrXHzEFbqWHQXeJxeS/view?usp=sharing
Pre-trained weight: https://drive.google.com/file/d/18Ni-P_gbMFkQb5pIbGn2AmOG8wHhQKTm/view?usp=sharing