The mutli-task work of <Monocular Outdoor Semantic Mapping with a Multi-task Network>
The repo is the multi-network part of
You can use python train --device 0 --dataset cs --ratio 0.85 to start your code journey
--device is to choose your GPU ID
--data is to choose your dataset, 'cs' or 'kitti'
--ratio is to choose the ratio between depth and semantic segmantation in loss function.
Enjoy