guilherme.schardong/facebias

Go to file

Guilherme Schardong 9985b91b75 Added the benchmarking script (Not working properly)

2026-05-29 15:52:49 +01:00

Adding some missing files from the last commits.

2026-05-29 15:51:30 +01:00

.gitignore

Adding some missing files from the last commits.

2026-05-29 15:51:30 +01:00

benchmark.py

Added the benchmarking script (Not working properly)

2026-05-29 15:52:49 +01:00

Makefile

Some auxiliary rules for make.

2026-04-22 13:31:02 +01:00

pyproject.toml

Added setup and basic module sources.

2026-04-15 22:24:58 +01:00

README.md

Adding some missing files from the last commits.

2026-05-29 15:51:30 +01:00

requirements.txt

Added pandas as dependency. More below.

2026-04-20 16:50:29 +01:00

README.md

ISR Face Dataset/Model Bias Check API

We use FairFace and MiVOLO version 1, face-only checkpoint.

Dataset

Download the tar file VISTEAM-NAS/Public_Data/facing2-skin-tone-train-images.tar.bz2 to the data directory, and extract it. This dataset has balanced sex and skin-tone, and unbalanced age.