This repository provides source code and trained models of the paper
"Coddora: CO2-Based Occupancy Detection Model Trained via Domain Randomization" to be presented at the International Joint Conference on Neural Networks (IJCNN).
Coddora is a simulation-trained deep learning model for detecting occupancy from CO2 rates in building rooms.
It is intended to support rapid application in practice, without the need for extensive collection of training data in real-world buildings.
The simulated dataset used for model training can be found here:
https://doi.org/10.5281/zenodo.10507614
An introduction on how to apply the Coddora model can be found here: HOWTO
When using the presented code or models please cite:
Manuel Weber, Farzan Banihashemi, Davor Stjelja, Peter Mandl, Ruben Mayer, and Hans-Arno Jacobsen. 2024. Coddora: CO2-Based Occupancy Detection Model Trained via Domain Randomization. In International Joint Conference on Neural Networks (IJCNN). June 30 - July 5, 2024, Yokohama, Japan.