This is the official code of TRG-Net, the proposal of our paper A Lightweight Gaussian-Based Model for Fast Detection and Classification of Moving Objects. TRG-Net is a unified model that can be executed on computationally limited devices to detect and classify only moving objects. This solution is based on the Faster R-CNN architecture, but with a novel GMM-based region proposal method.

Install with:

$ pip install -r requirements.txt
$ python setup.py install

Paste the pre-trained model in the following route: ~/.trgnet/checkpoints/, feel free to send us an email for the .pt file. If you want to train the model by yourself run the train.py file. TRG-Net uses the Kitti dataset, the dataset will be automatically downloaded if not present.

Finally, run the sample.py script to run the model and start detecting moving objects.

@conference{visapp23trgnet,
  author={Joaquin Palma{-}Ugarte. and Laura Estacio{-}Cerquin. and Victor Flores{-}Benites. and Rensso Mora{-}Colque.},
  title={A Lightweight Gaussian-Based Model for Fast Detection and Classification of Moving Objects},
  booktitle={Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 5: VISAPP,},
  year={2023},
  pages={173-184},
  publisher={SciTePress},
  organization={INSTICC},
  doi={10.5220/0011697200003417},
  isbn={978-989-758-634-7},
  issn={2184-4321},
}