search

JoyceWLee / LDCTIQAG_reference_docker

3 stars 1 forks source link

About the submission code #1

Open Houruizhi opened 1 year ago

Houruizhi commented 1 year ago

In line 49 in the process.py file, the saved result is self._case_results[0]. However, if there are series of images in the testing dir, the result self._case_results is like: [[2.8333334922790527], [2.0], [3.0], [2.1666667461395264], [3.0], [2.0], [2.1666667461395264], [1.8333333730697632], [1.6666667461395264], [2.8333334922790527]].

I wonder what's the directory structure of the official testing dir /input/images/synthetic-ct/?

Thanks

JoyceWLee commented 1 year ago

In the testing directory /input/images/synthetic-ct/, there will be test batches, each consisting of 100 stacked images (i.e., shape=(100, 512, 512)). Within the Grand Challenge platform, the algorithm is executed for each test batch, and a JSON file will be saved for each batch. Therefore, your output JSON file must contain a list of floats corresponding to one test batch. Otherwise, the file will not be ready for evaluation.

Given that there will be only one list in the output JSON file, saving the first item in self._case_results won't be a problem during actual testing.