search

JStehouwer / FFD_CVPR2020

108 stars 22 forks source link

Use pretrained model to classify own data? #12

Open jtlz2 opened 3 years ago

jtlz2 commented 3 years ago

Hi @JStehouwer - thank you so much for the awesome code (v2.1)!

I am trying to use your pretrained model on my own images in order to try out the classifier.

Are you able to confirm:

Thanks again

JStehouwer commented 3 years ago

Thanks for your interest in our work!

In the releases tab of this Github Repo, you should be able to find trained models for Xception with Regression Map and with MAM map. The test.py file demonstrates how these models can be loaded and evaluated on a dataset of images.

Let me know if you need anything else or if you aren't able to get it working.

mrybak-Novetta commented 3 years ago

I'm trying to get these working now. I saw in the paper that faces were detected with InsightFace. How do you preprocess the detections before feeding them to the model? I can't find it in the code anywhere.

mrybak-Novetta commented 3 years ago

Ah, i see the relevant info is in the scripts that accompany the dataset.

lloyd31 commented 2 years ago

Hi @JStehouwer , @mrybak-Novetta

I am trying to run the model over my own dataset as well. I am trying to use the classifier over a given set of fake images. However i am a bit stuck in terms of generating my maps without the real image? Could you please suggest me on how to proceed further?