samleoqh / MSCG-Net

Multi-view Self-Constructing Graph Convolutional Networks with Adaptive Class Weighting Loss for Semantic Segmentation
MIT License
68 stars 28 forks source link

test #35

Closed zwsbh closed 4 months ago

zwsbh commented 4 months ago

How to display the test results, which is the confusion matrix

samleoqh commented 4 months ago

pls refer to lib/utils/measure.py

zwsbh commented 4 months ago

There is no the results of the confusion matrix when running text_submission. py

samleoqh commented 4 months ago

test_submission is just for the hold-out test, not for validation, there is no ground-truth for the hold-out test set, we only submit the results to the Challenge organizations, and then they will evaluate your submission on their hold-out ground-truth.

zwsbh commented 4 months ago

Is the indicator in the paper the result of the validation set or the test set

zwsbh commented 4 months ago

How to obtain the results of comparative experiments without real labels from the test set

samleoqh commented 4 months ago

Just compare models on your validation set. You cannot get the IOU or F1 scores on the hold-out test set now since the workshop has already ended and closed any submission evaluations.