search

wvangansbeke / Unsupervised-Classification

SCAN: Learning to Classify Images without Labels, incl. SimCLR. [ECCV 2020]
https://arxiv.org/abs/2005.12320
Other
1.37k stars 268 forks source link

eval can't run the result #117

Closed kkellyk closed 2 years ago

kkellyk commented 2 years ago

Hi, I run the !python eval.py --config_exp configs/scan/scan_cifar10.yml --model repository_eccv/cifar-10/scan/model.pth.tar --visualize_prototypes but but no result:

Load model weights ... Perform evaluation of the clustering model (setup=scan). Figure(640x480) {'ACC': 0.8142, 'ARI': 0.658409853352927, 'NMI': 0.7090852722819195, 'ACC Top-5': 0.9885, 'hungarian_match': [(0, 8), (1, 7), (2, 0), (3, 6), (4, 9), (5, 3), (6, 5), (7, 2), (8, 1), (9, 4)]} Get topk Figure(640x480) Figure(640x480) Figure(640x480) Figure(640x480) Figure(640x480) Figure(640x480) Figure(640x480) Figure(640x480) Figure(640x480) Figure(640x480)

I added %matplotlib inline at the top of visualize_indices function in the eval.py file. But the result is still the same.

How can I do? Thanks Kelly

emessori commented 1 year ago

Hello, could you please share the solution if you solved the issue? Thank you and have a great day.