Activity recognition in FPV using multi-modal data
This is a repository for training the deep learning model used in ANGEL system. The repository has majorly two components (as folders) - activity_hydra
and mmdetection3d
:
activity_hydra
consists of the boilerplate framework for training and inference of neural network architectures used in ANGEL systems, generated from Pytorch Lightning + Hydra frameworks.mmdetection3d
is the implementation of Open MMDetection 3D library with modifications to accomodate the support of H2O dataset.
The installation and running instructions are mentioned in the README availabe with each of these folder. For more information of each of the frameworks/libraries used in the above repository, refer to the following links: