search

SarahwXU / HiSup

MIT License
131 stars 20 forks source link

Unable to Reproduce the results presensted in the paper #29

Open nsawant-ict opened 5 months ago

nsawant-ict commented 5 months ago

Hello Authors,

Thank you for this amazing work. I am trying to reproduce the results that you have mentioned in the paper. I ran the following command

python scripts/test.py --config-file config-files/crowdai_hrnet48.yaml

And the results were as follows: Screenshot from 2024-06-24 17-08-36

As you can see the results are very far as compared to the ones mentioned in the paper. I am wondering what could have happened? If I am doing something wrong while running the test.py script then please let me know. I am looking forward for some help on this.

SarahwXU commented 4 months ago

Hi, thanks for your attention. Did you download the provided pretrained models for AICrowd dataset? The evaluation scores seem too low. One guess is that the well-trained parameters may not loaded. We are happy to help if you could give more details.

nsawant-ict commented 4 months ago

Hey, yes I have downloaded the AICrowd dataset. Could you let me know what specific details are you looking for?

SarahwXU commented 2 months ago

Please check if the pretrained parameters are loaded successfully.