ViPTT-Net: Video pretraining of spatio-temporal model for tuberculosis type classification from chest CT scans
ViPTT-Net is a method that pretrains a hybrid CNN-RNN based model on realistic videos for human activity recognition task. It is then fine-tuned on a dataset of chest CT scans for the task of tuberculosis type classification.
ViPTT-Net achieved 2nd place (Kappa score of 0.2) in the ImageCLEF 2021 Tuberculosis - TBT Classification Challenge.
Figure 1. Schematic layout of ViPTT-Net.
- Paper (Published, arXiv)
- Task details are here
- Leaderboard results
If you use this code or models in your scientific work, please cite the following paper:
H. Zunair, A. Rahman, N. Mohammed, ViPTT-Net: Video pretraining of spatio-temporal
model for tuberculosis type classification from chest CT scans, in: CLEF2021 Working
Notes, CEUR Workshop Proceedings, CEUR-WS.org <http://ceur-ws.org>, Bucharest,
Romania, 2021.
This code requires:
- Python 3.7
- TensorFlow 2.4.1
- Nibabel
This research code will not be maintained, unless we decide to do a follow up work. If you have trouble running this code ONLY with the requirements mentioned above, file and issue and we'll look at it tomorrow.
See notebooks/
.
See notebooks/
.
See notebooks/
.
We provide pretrained models:
Models | Description | Weights |
---|---|---|
ViPTT-Net ImageCLEF | Fine-tunes ViPTT-Net UCF50 on ImageCLEF 2021 Tuberculosis - TBT dataset. |
ViPTT-Net-CLEF-TBT.h5 |
ViPTT-Net UCF50 | Trains ViPTT-Net on a subset of the UCF50 dataset | ViPTT-Net-UCF50.h5 |
See paper for details!
MIT