Clone the repo
git clone https://github.com/MatthewHowe/WIBAM.git
cd WIBAM
Download the demo dataset from Zenodo. Put it in the WIBAM directory under data - as follows, rename it to wibam.
WIBAM
│ README.md
│ requirements.txt
│ ...
|
└───data
│ └───wibam
| └───calib
│ └───annotations
│ └───frames
| | └───0
| | └───1
| | └───2
| | └───3
│ └───image_sets
| └───models
│
└───src
└───lib
└───tools
| ...
Pull the docker image
docker pull matthewhowe/wibam
Run the docker image
make run
From within the docker image run
python src/main_lit.py ddd --dataset=wibam --load_model=data/wibam/models/wibam.ckpt --batch_size=1 --save_video --gpus=0 --num_workers=1 --test_only
Three videos should output: The input image with predictions + ground truth overlay, a BEV, and the two joint. All detections and prediction statistics will be generated under csv_results/{model_name}.csv