Skip to content

Latest commit

 

History

History
71 lines (53 loc) · 2.13 KB

README.md

File metadata and controls

71 lines (53 loc) · 2.13 KB
Raw

DataParser for CORSMAL Containers Manipulation

Python script for parsing the CORSMAL Containers Manipulation dataset, including train set, public test set, and private test set. The script includes Numpy, OpenCV, and PyTorch libraries for testing and starting the development of solutions to solve different tasks on the dataset. To simply test the parsing of the dataset, libraries can be commented out.

Dataset

Tested on

  • Python 3.8.5
  • Server machine with CentOS Linux release 7.7.1908

Installation

Download or clone the repository.

git clone https://github.com/CORSMAL/CCM-DataParser.git

How to run

Use the following command and modify the options accordingly:

python ccm_data_parser.py --datapath <DATAPATH> --set [train, pubtest, privtest]

Starting kit

If you wish to integrate your software directly into the current script, we suggest the following libraries (already tested):

  • OpenCV 4.4.0
  • PyTorch 1.6.0
  • TorchVision 0.7.0
  • NVIDIA CUDA 10.1
  • Anaconda3 (4.7.12)
  • Numpy 19.1.1

You can find some commands in the script that are commented to start integrating the software in the script.

To overcome root privilege issues and favour reproducibility across machine with different compatibilities, we recommend to create virtual environments and we suggest Anaconda/Miniconda. See the following tutorial for more infor on how to install miniconda: Miniconda tutorial

conda create -n CCM
source activate CCM

conda install scipy numpy
conda install -c conda-forge opencv
conda install -c pytorch torchvision
pip install pickle5

Pickle library is necessary for reading calibration files.

The script will output Python and OpenCV versions and if GPU is enabled.

Enquiries, Question and Comments

If you have any further enquiries, question, or comments, please contact [email protected] If you would like to file a bug report or a feature request, use the Github issue tracker.

Licence

This work is licensed under the MIT License. To view a copy of this license, see LICENSE.