Project folder for Elective in AI 1, held by prof. Fiora Pirri at Sapienza Università di Roma. Developed by Damiano Zappia
This project has the aim of implement a code that utilize the Keras Retinanet with 2 different datasets.
- First assignment is to run the network on one of the default datasets and find the weights
- Second assignment is to run the network on the AVA dataset and find the error
In the report pdf file you can find a detailed description of the work done, with explainations on the procedure, the implementation choices and the results
Drive links to the heavy model files are provided below:
- Keras Retinanet trained on pascal VOC2007 - 14 epochs : https://drive.google.com/file/d/1mWJRmjI56Rg_ev7mRQCGu6SPXG80_oXS/view?usp=sharing