0.get_started.md

Get Started

Installation

1. clone this repo.

   git clone https://gitlab.phigent.io/xianda.guo/openstereo
   

2. Install dependenices:

   • pytorch >= 1.13.1
   • torchvision
   • pyyaml
   • tensorboard
   • opencv-python
   • tqdm

   Install dependenices by Anaconda:

   conda install tqdm pyyaml tensorboard opencv -c conda-forge
   conda install pytorch torchvision torchaudio pytorch-cuda=11.6 -c pytorch -c nvidia
   

   Or, Install dependenices by pip:

   pip install tqdm pyyaml tensorboard opencv-python
   pip install torch torchvision
   

Prepare dataset

See prepare dataset.

Get trained model

Go to the model zoom, then download the model file and uncompress it to output.

Train

Train a model by

CUDA_VISIBLE_DEVICES=0,1 python openstereo/main.py --config ./configs/psmnet/PSMNet_sceneflow.yaml --scope train

• --config The path to config file.
• --scope Specified as train, val or test_kitti.
• --restore_hint You can specify a number of iterations or use restore_hint in the config file and resume training from there.
• --master_addr The master address of DDP.
• --master_port The master port of DDP.
• --no_distribute If specified, the program will not use DDP to train.
• --device The device to use, e.g. cuda:0. only used when no_distribute is specified.

Val

Evaluate the trained model by

CUDA_VISIBLE_DEVICES=0,1 python openstereo/main.py --config ./configs/psmnet/PSMNet_sceneflow.yaml --scope val

• --phase Specified as test.
• --restore_hint Specify a checkpoint.

Tip: Other arguments are the same as train phase.

Customize

1. Read the detailed config to figure out the usage of needed setting items;
2. See how to create your model;
3. There are some advanced usages, refer to advanced usages, please.