sjtu-openmmlab-tutorial

This repository maintains codes of OpenMMLab tutorial delivered in SJTU. Tutorial codes of OpenMMLab version 1.0 and 2.0 are organized in v1.0 and v2.0 respectively.

For a full list of OpenMMLab courses and tutorials, see OpenMMLabCourse (in Chinese).

Notes on SJTU Cluster

For OpenMMLab v2.0

See this

For OpenMMLab v1.0

1. To activate conda environment in terminal

   module load anaconda3/2019.07
   source activate openmmlab

2. Remember to select openmmlab kernel when using jupyter notebook

3. Remember to check PATH in terminal

   PATH should include /cluster/apps/anaconda3/2019.07/envs/openmmlab/bin at the first entry. Otherwise mim will not work correctly.

Set up Environment Locally

Step 0. Set up Python and install PyTorch correctly, using either pip or conda

Step 1. Install MIM

pip install openmim
which mim  # to check mim installation
# On windows, in Powershell, use
# gcm mim

For OpenMMLab v2.0

Step 2. Install MMCV

mim install "mmcv>=2.0.0rc0"

Step 2. Install MMClassification and MMDetection

mim install "mmcls>=1.0.0rc0" "mmdet>=3.0.0rc0"

For OpenMMLab v1.0

Step 2. Install MMCV

mim install mmcv-full

Step 3. Install MMClassification and MMDetection

mim install mmcls mmdet

License

Apache-2.0 license

Credit

This repository uses some datasets and models from Zihao](https://github.com/TommyZihao/MMClassification_Tutorials).

Fruit dataset

Fruit model

and MMDetection and KITTI.

KITTI-tiny