PyTorch-Doubtlab

The Doubtlab library modified to work with PyTorch models

At the moment, the four reasons below are available:

• ProbaReason: assign doubt when a models' confidence-values are low for any label
• WrongPredictionReason: assign doubt when a model cannot predict the listed label
• ShortConfidenceReason: assign doubt when the correct label gains too little confidence
• LongConfidenceReason: assign doubt when a wrong label gains too much confidence

Dataset Structure

The scripts expect the data to be arranged in the following structure:

Dataset
├ train
| ├ class 1
| | ├ image_1.png
| | └ image_2.png
| ├ class 2
| └ class 3
├ valid
| ├ class 1
| ├ class 2
| └ class 3
└ test
| ├ class 1
| ├ class 2
| └ class 3

Usage

To generate potentially mislabeled images, you will need to:

1. Train a torchvision model on your data
2. Use your model with doubtlab to identify mislabels

Training: Run the Torch_Image_Classification_Training script, following the dataset structure above. Modify the number of classes before loading the model.

Doubtlab: Follow the example doubtlab usage in Torch_Doubtlab_Example