The paper is accpeted to ICRA2022.

• The model uses the SEQS-05 Sythia-seq collections here
• Using different collections might require modifications to the dataloader. Please check the specific data structure and labels.
• Extract the zip / tar and modify the path appropriately in your ./configs/deeplab_synthia.yml

path: /datasets/synthia-seq/

conda env create -f requirements.yaml
conda activate recallCE

• Run the training script.

python train.py --config ./configs/deeplab_synthia.yaml

1. For testing on Sythia dataset, change the validation split to test.

val:
    split: test

2. Run the testing script

• Qulitative results will be saved in runs/synthia/rgbd_synthia

python test.py --config ./configs/deeplab_synthia.yaml

• This work was supported by ONR grant N00014-18-1-2829.
• This code is built upon the implementation from Pytorch-semseg.