We propose our dataloader API for loading and visualizing our RoboMaster opensource datasets. The datasets includes:
- Sentry detection data DRL-CASIA/MVM3D
- Vehicle mounted detection data DRL-CASIA/ABCP, DRL-CASIA/Detection-Datasets-for-ABCP.
- Vehicle mounted depth estimation data (In progress).
Note that this API is only tested on ubuntu/Mac devices which has:
- python 3.6+
- torch 1.7.1
- torchvision 0.8.2
- ...
First, clone this repo:
git clone [email protected]:DRL-CASIA/NeuronsDataset.git
cd NeuronsDataset
Then, pack the code into wheel packages for later installation:
conda activate YourEnv # Activate your python environment first.
cd src
pip3 setup.py bdist_wheel
The commands above are expected to generate a whl file and several folders.
Next, install the whl package into your python3 environment using pip:
cd dist
pip3 install neuronsdataset-1.0-py3-none-any.whl
The generated folders could be safely deleted after installing the API.
Please refer to main.py for details.
We also provide samples of dataset mentioned in Introduction section, which are contaied in sample_data.