The files in this code base can be used to replicate the experiments and results in report.pdf
This python notebook contains the implementation of the JLU training algorithm.
This python notebook contains the implementation to load model variable and prepare tsv file for t-SNE embeddings.
This python notebook contains the implementation to load a model and test it on specific datasets.
- gen_list.csv: For the dataset OD, gender labels. 1: MALE, 0: FEMALE (always this notation is used)
- age_list.csv: For the dataset OD, age labels.
- gen_list_balanced.csv: For the dataset EBD, gender labels.
- age_list_balanced.csv: For the dataset EBD, age labels.
- gen_list_extreme_biased.csv: For the dataset BD, gender labels.
- age_list_extreme_biased.csv: For the dataset BD, age labels.
The final data and CSV files can be directly used from SportsDataScraper/SportsDataScraper/img_download/FinalData/. If you wish to replay the scraping, run the individual spiders and pipe it to the appropriate files. Next, run the image cleaners and resizers for each data source followed by the face_prediction scripts. You will need to have a microsoft account and change the APItoken with your generated token.