MIC-DKFZ / medicaldetectiontoolkit

The Medical Detection Toolkit contains 2D + 3D implementations of prevalent object detectors such as Mask R-CNN, Retina Net, Retina U-Net, as well as a training and inference framework focused on dealing with medical images.
Apache License 2.0
1.31k stars 297 forks source link

How can I only evaluate based on the ground-truth bounding box? #128

Open sunformoon opened 3 years ago

sunformoon commented 3 years ago

Thanks for the repo! I am wondering how can I get the evaluation performance (mainly the classification results) given the ground-truth bounding box?

sunformoon commented 3 years ago

To clarify, my major task is the classification task, so I would like to obtain the classification results based on the given ground-truth bounding boxes. However, I trained with the detection + classification + segmentation label, in which case I have much more voxel-level annotation to prevent the model from overfitting since I don't have many CT cases.

@pfjaeger Any comments would be greatly appreciated!