MediaBrain-SJTU / MVFA-AD

[CVPR2024 Highlight] Adapting Visual-Language Models for Generalizable Anomaly Detection in Medical Images
MIT License
110 stars 16 forks source link

Question about 'test' function while computing img-AUROC for Dataset Brain #9

Open fuyansheng opened 3 months ago

fuyansheng commented 3 months ago

In the function 'test' of test_few.py(as well train_few.py), the current code computes img-AUROC of the dataset Brain (as well as Liver, Retina_RESC) using the _seg_patchtokens, instead of _det_patchtokens, which is different with the paper. Will it make differences in testing result?