ttanida / rgrg

Code for the CVPR paper "Interactive and Explainable Region-guided Radiology Report Generation"
MIT License
146 stars 25 forks source link

Is there seperate weights for Faster R-CNN + Binary classifiers? #25

Closed YatingPan closed 8 months ago

YatingPan commented 8 months ago

Hi, I want to use the Faster R-CNN and binary classifiers trained on Chest ImaGenome, but didn't find the isolated weights for these components. I extracted the weights for these 2 components from the full model weights, but based on some inference on Chest ImaGenome's test datasets, the binary classifiers' outputs are not so accurate... Are there any isolated weights or evaluation results for this part? Thank you very much!

ttanida commented 8 months ago

Hi,

please take a look at this issue for the object detector weights: https://github.com/ttanida/rgrg/issues/3 and this issue for the object detector + binary classifier weights: https://github.com/ttanida/rgrg/issues/10

Best, Tim

Am Mi., 20. März 2024 um 11:02 Uhr schrieb Yating Pan < @.***>:

Hi, I want to use the Faster R-CNN and binary classifiers trained on Chest ImaGenome, but didn't find the isolated weights for these components. I extracted the weights for these 2 components from the full model weights, but based on some inference on Chest ImaGenome's test datasets, the binary classifiers' outputs are not so accurate... Are there any isolated weights or evaluation results for this part? Thank you very much!

— Reply to this email directly, view it on GitHub https://github.com/ttanida/rgrg/issues/25, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI7Q3AKCFOWZJ3AWCPSZWNTYZFNEVAVCNFSM6AAAAABE7EUDLWVHI2DSMVQWIX3LMV43ASLTON2WKOZSGE4TOMJTGEZTQMI . You are receiving this because you are subscribed to this thread.Message ID: @.***>

YatingPan commented 8 months ago

Ok Thanks!