Project carried out for the Computer Vision exam. The implemented system provides a tool for locating and classifying famous brand logos within images Thanks to this project, it is possible to achieve obscuration of logos within movies without the need to locate them manually, and also gives the possibility to choose which logos to accept and which not through a whitelist.
The code was divided into two .ipynb notebooks. The notebook Computer_Vision_Project.ipynb contains the code related to the creation of the architecture, training, and evaluation of the produced models. Since the notebook requires the presence of the datasets within the same folder, the latter cannot be executed, but outputs are left to be viewed by the reader. The App.ipynb notebook contains code related to the creation of an application that demonstrates concrete uses of the models and code produced in the previous notebook. In this case it is possible to execute the code, should the necessary data be downloaded through the appropriate cells. The application produced is intended to be run at the Google Colaboratory portal; execution in different environments may result in errors. In addition to the code, there is also a presentation in pdf format, in which the most important aspects of the implemented system are described.