Privacy-aware, low-power, and device-free indoor occupancy detection is highly sought after in order to enable smart environments. IR-based solutions seem to obey to these demands, however, are quite limited in terms of real-life application due to several obstacles. So, another, more appropriate solution, is to be found. For this, Ranga Rao and Sujay from the Networked and Embedded Systems group of the TU Delft already developed a device that integrates, among others, a thermophile sensor. For me, it is now the task to explore the sensor usage possibilities. This is done in three steps:
- Acquiring sensor data appropriately
- Applying computer vision techniques to detect people based on IR information
- Sending information extracted from the sensor, using the MQTT protocol, to an Azure IoT Hub.
For the MQTT connection with the IoT Hub, also a separate code is created for testing and can be found in the folder standalone_MQTT_IoT
Hardware used in the project:
-
Computer
-
LOCI device housing thermopile and PIR sensors (see image)
Languages, libraries and versions used in the project:
- Python 3.8 with libs asyncio, math, struct, and datetime
- Serial 3.5
- Numpy 1.22
- Matplotlib 3.4.2
- OpenCV 4.5.5
- Azure IoT Python SDK v2
Since this is a development project, no command line UI was added, the code is run simply by one line:
$ python project_main.py
When the code is run without any adjustments it will default to detecting people in one frame that is given by the exemplary data. This, however, can be changed by adjusting three constants at the beginning of the project_main.py file:
- LIVE
- VID
- CONTOUR
The LIVE constant can be set to True if it is desired to deal with live serial data from the LOCI device The VID constant can be set to True if it is desired to show an entire sequence of frames in which people are searched for. This only works when the LIVE equals True. The CONTOUR constant can be set to False if it is not desired that contours are shown around the detect people in a frame
Kilian van Berlo - S5436737 - [email protected]
Project Link: https://github.com/KilianBerlo/Privacy-Aware-Device-Free-Occupancy-Detection