v1 (MVA paper)

This release contains code for reproducing results from the paper "A baseline for semi-supervised learning of efficient semantic segmentation models".

Pre-trained DeepLab v2 parameters

Below are pre-trained parameters for DeepLab v2 with ResNet-101 as backbone that have been converted according to the procedure described here by @kazuto1011:
deeplabv1_resnet101-coco.pth