ImgRec
This read me file describes the steps to run the code for "End-to-End Image-Based Recommender System" paper.
Requirements* 1- numpy 2- pandas 3- tensorflow 2.3.0 4- scikit-learn
To run any of the scripts it is required to check the "data path" of the required file, then running command is ==> python file_name e.g. "python imgrec_amazon_fashion_ResNetFeatures_FT.py"
For ImgRec-EtE, we split the training process into two phases to save time and processing power; Phase 1: fixing the image network parameters. Phase 2: jointly learning the last 50 layers of the image network.
To run the experimemts firstly run the "imgrec_men_finetuned_SaveModel.py", which will run with fixed pre-trained parameters. Then run the "imgrec_men_finetuned_LoadModel.py" to obtain the paper's results.