Question for validation result. #4
Description
Excellent work!!! Thanks for sharing the code! But when I tried to run the code(train.py and train_envelope.py), I met a problem. As shown in the figure, the val results lost the final class (terrain in step_0, traffic-sign in step_1, car in step_2) in every step. And the "Average (mIoU)' seems to compute all 19 classes.
