Evaluation container for the MIDOG 2021 challenge on grand-challenge.org. This container is losely based on the evalutils-created evaluation container from grand-challenge.org, yet is utilizes new (so far undocumented) interfaces.
This container is provided for transparency reasons and to improve clarity of evaluation for the participants.
The grand-challenge.org system provides a file /input/predictions.json to the container, esentially consisting of a list of dictionaries where each dictionary contains the results of a single image (and thus a single docker container run).
The container provides detailed outputs for each image as well as the overall F1 score, precision and recall.
{ "case": { "007.tiff": { "true_positives": 2, "false_negatives": 0, "false_positives": 1 } }, "aggregates": { "precision": 0.6666666666666666, "recall": 1.0, "f1_score": 0.8 } }