Image classification tasks via CNN models, leveraging Monte Carlo dropout during inference to assess model uncertainty
Related paper: [Gal et al., Dropout as a bayesian estimation: representing model uncertainty in deep learning, ICML 2016
- Generate small image tiles from larger acquisitions
- Split dataset into training, validation and test sets, using stratification strategy at acquisition level
- Neural network training
- Neural network inference (direct)
- Neural network inference with gradCAM visualization
- Neural network inference with uncertainty assessment
Use of ResNet models
Use of fastai_v1 (2019)