We take CSRNet as backbone to develop our framework on ShanghaiTechRGBD benchmark.
We strongly recommend Anaconda as the environment.
Python: 2.7
PyTorch: 0.4.0
Generation the ground-truth density maps for training (please edit the dataset root path in the script).
python RGBD_GT_generation.py
Make data path files and edit this file to change the path to your original datasets.
python make_json.py
Edit this file for training CSRNet-based IADM model.
bash train.sh
Edit this file for testing models.
bash test.sh
You can also test our checkpoints released at here