Logistic Activity Recognition Challenge (LARa)

Implementation code for the Annotation Tool that is used for LARa dataset, presented in the Journal "LARa: Creating a Dataset for Human Activity Recognition in Logistics Using Semantic Attributes", see https://www.mdpi.com/1424-8220/20/15/4083

And

ICPR20:From Human Pose to On Body Devices for Human Activity Recognition

Implementation code for "From Human Pose to On Body Devices for Human Activity Recognition, see https://ieeexplore.ieee.org/document/9412283

Prerequisites

The implementation is done in Python 3.7:

-numpy

-scipy

-scikit-learn

-PyQt5

-pygtgraph

-pytorch

-PyOpenGL

-dill

Dataset

LARa dataset can be downloaded in https://zenodo.org/record/3862782#.XtVJOy9h3UI

Networks

Networks are available in: https://tu-dortmund.sciebo.de/s/YkpqlYOffFrmFr0

Place the networks in a folder called "networks"

Example

Running the main.py script in Annotation_Tool_LARa.

For using the tCNNs for predicting activities classes, download the 'class_network.pt' and 'attr_network.pt' from LARa dataset.
Store the networks 'class_network.pt' and 'attr_network.pt' in Annotation_Tool_LARa/networks/ Annotation_Tool_LARa/networks/class_network.pt Annotation_Tool_LARa/networks/attr_network.pt

Contact

Erik Altermann erik.altermann@tu-dortmund.de
Fernando Moya Rueda fernando.moya@tu-dortmund.de

Technische University of Dortmund Department of Computer Science Dortmund, Germany

The work on this publication was supported by Deutsche Forschungsgemeinschaft (DFG) in the context of the project Fi799/10-2, HO2403/14-2 ''Transfer Learning for Human Activity Recognition in Logistics''.

jia-wei-zheng / Annotation_Tool_LARa