Optimisation of a Raspberry Pi-based Bioacoustic Sensor for Data Collection

Joel Muhanguzi, Gabriel Kiarie, Ciira Maina, Ernest Mwebaze
DLI 2025 Research Track, PMLR 302:1-9, 2026.

Abstract

This paper details the optimisation of a Raspberry Pi-based bioacoustic sensing system designed for data collection for acoustic classification of birds. Optimised sensors are desirable when collecting data off-the-grid. The acoustic sensor’s power consumption was studied under various conditions, such as using a full and Lite Raspberry Pi OS to guide the sensor’s optimisation. The power management board used to power the sensor was also redesigned to improve efficiency. A machine learning model to classify 3 bird species and other sounds with an accuracy of 93% and loss of 30% on the test set was developed. An 80%-10%-10% train, test and validation split ratio was used to train and evaluate the model which was then quantized to tensorflowlite to run directly on the raspberry pi hardware in a miniaturized format. The optimised sensor was then deployed at a Wildlife Conservancy in Africa, Kenya to continously collect data and perform inference at the edge. Keywords: Bioacoustics, Raspberry Pi, Sensor Optimisation, Off-grid Data Collection, Power Management, machine learning, quantization, miniaturized.

Cite this Paper

BibTeX
@InProceedings{pmlr-v302-muhanguzi26a, title = {Optimisation of a Raspberry Pi-based Bioacoustic Sensor for Data Collection}, author = {Muhanguzi, Joel and Kiarie, Gabriel and Maina, Ciira and Mwebaze, Ernest}, booktitle = {DLI 2025 Research Track}, pages = {1--9}, year = {2026}, editor = {Haddad, Hatem and Kahira, Albert Njoroge and Bourhim, Sofia and Olatunji, Iyiola Emmanuel and Makhafola, Lesego and Mwase, Christine}, volume = {302}, series = {Proceedings of Machine Learning Research}, month = {17--22 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v302/main/assets/muhanguzi26a/muhanguzi26a.pdf}, url = {https://proceedings.mlr.press/v302/muhanguzi26a.html}, abstract = {This paper details the optimisation of a Raspberry Pi-based bioacoustic sensing system designed for data collection for acoustic classification of birds. Optimised sensors are desirable when collecting data off-the-grid. The acoustic sensor’s power consumption was studied under various conditions, such as using a full and Lite Raspberry Pi OS to guide the sensor’s optimisation. The power management board used to power the sensor was also redesigned to improve efficiency. A machine learning model to classify 3 bird species and other sounds with an accuracy of 93% and loss of 30% on the test set was developed. An 80%-10%-10% train, test and validation split ratio was used to train and evaluate the model which was then quantized to tensorflowlite to run directly on the raspberry pi hardware in a miniaturized format. The optimised sensor was then deployed at a Wildlife Conservancy in Africa, Kenya to continously collect data and perform inference at the edge. Keywords: Bioacoustics, Raspberry Pi, Sensor Optimisation, Off-grid Data Collection, Power Management, machine learning, quantization, miniaturized.} }
Endnote
%0 Conference Paper %T Optimisation of a Raspberry Pi-based Bioacoustic Sensor for Data Collection %A Joel Muhanguzi %A Gabriel Kiarie %A Ciira Maina %A Ernest Mwebaze %B DLI 2025 Research Track %C Proceedings of Machine Learning Research %D 2026 %E Hatem Haddad %E Albert Njoroge Kahira %E Sofia Bourhim %E Iyiola Emmanuel Olatunji %E Lesego Makhafola %E Christine Mwase %F pmlr-v302-muhanguzi26a %I PMLR %P 1--9 %U https://proceedings.mlr.press/v302/muhanguzi26a.html %V 302 %X This paper details the optimisation of a Raspberry Pi-based bioacoustic sensing system designed for data collection for acoustic classification of birds. Optimised sensors are desirable when collecting data off-the-grid. The acoustic sensor’s power consumption was studied under various conditions, such as using a full and Lite Raspberry Pi OS to guide the sensor’s optimisation. The power management board used to power the sensor was also redesigned to improve efficiency. A machine learning model to classify 3 bird species and other sounds with an accuracy of 93% and loss of 30% on the test set was developed. An 80%-10%-10% train, test and validation split ratio was used to train and evaluate the model which was then quantized to tensorflowlite to run directly on the raspberry pi hardware in a miniaturized format. The optimised sensor was then deployed at a Wildlife Conservancy in Africa, Kenya to continously collect data and perform inference at the edge. Keywords: Bioacoustics, Raspberry Pi, Sensor Optimisation, Off-grid Data Collection, Power Management, machine learning, quantization, miniaturized.
APA
Muhanguzi, J., Kiarie, G., Maina, C. & Mwebaze, E.. (2026). Optimisation of a Raspberry Pi-based Bioacoustic Sensor for Data Collection. DLI 2025 Research Track, in Proceedings of Machine Learning Research 302:1-9 Available from https://proceedings.mlr.press/v302/muhanguzi26a.html.

Related Material