keywords: Deep learning, Audio analysis, smart bee-hive monitoring, precision beekeeping, feature selection, embedded system
Bees are very important pollinating insects contributing to preserve natural ecosystems. How-
ever, they are also sensitive to various external factors such as weather, diseases, predators or pollution which can have severe impacts on their health. This explain the recent researches based on IA to develop smart beehive monitoring systems to assist beekeepers. Recently, the acoustic analysis approach for precision beekeeping gained interest due to the capability of audio signal to convey accurate information about the health state of a beehive using a simple microphone (e.g. the number of bees, stress factors, the absence of the queen, etc.). Hence, estimating relevant information from audio signals requires robust acoustic features and the adequate preprocessing (e.g. signal separation and denoising) which could lead to promising result when combined with a deep learning approach. Moreover, the usage of an embedded system introduces constraints about the computational cost and the amount of transmitted data that should be optimized to be as low as possible.
The goal of this PhD thesis is to design a complete method based on deep learning allowing to collect data and to efficiently predict the state of a beehive using an embedded measurement system in a real-world field recording scenario.
The objectives of this thesis can be summarized as follows.
• Identifying the most efficient and robust audio features for supervised, and non-supervised audio classification scenarios.
• Development and comparative assessment of new deep learning methods for identifying a beehive health state from recorded audio signals.
• Optimal pre-processing and denoising to enhance the audio signal of interest (e.g. audio segmentation and event classification).
• Design of a complete solution based on an embedded system allowing to capture signal and to predict the state of a beehive.
• good machine learning and signal processing knowledges
• mathematical understanding of the formal background
• excellent programming skills (Python, Matlab, C++)
• good motivation, high productivity and methodical works
• an interest for AI and new technologies
 Stefania Cecchi, Alessandro Terenzi, Simone Orcioni, and Francesco Piazza. Analysis of the sound emitted by honey bees in a beehive. In Audio Engineering Society Convention 147, Oct 2019.
 Tymoteusz Cejrowski, Julian Szymanski, Higinio Mora, and David Gil. Detection of the Bee Queen Presence Using Sound Analysis. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), volume 10752 LNAI, pages 297–306, 2018.
 J Stephen Downie. Music information retrieval. Annual review of information science and technology, 37(1):295–340, 2003.
 Vladimir Kulyukin, Sarbajit Mukherjee, and Prakhar Amlathe. Toward Audio Beehive Monitoring: Deep Learning vs. Standard Machine Learning in Classifying Beehive Audio Samples. Applied Sciences, 8(9):1573, September 2018.
 MV Lima, JPAF De Queiroz, LAF Pascoal, EP Saraiva, KO Soares, and A Evangelista-Rodrigues. Smartphone-based sound level meter application for monitoring thermal comfort of honeybees apis mellifera l. Biological Rhythm Research, pages 1–14, 2019.
 Inaas Nolasco, Alessandro Terenzi, Stefania Cecchi, Simone Orcioni, Helen L. Bear, and Emmanouil Benetos. Audio-based identification of beehive states. arXiv:1811.06330 [cs, eess], November 2018. arXiv: 1811.06330.
 Antonio Robles-Guerrero, Tonatiuh Saucedo-Anaya, Efr ́en Gonz`alez-Ramerez, and Carlos E Galvan-Tejada. Frequency Analysis of Honey Bee Buzz for Automatic Recognition of Health Status: A Preliminary Study. Research in Computing Science, 142:89–98, 2017.
 Antonio Robles-Guerrero, Tonatiuh Saucedo-Anaya, Efr ́en Gonz ́alez-Ram ́ırez, and Jos ́e Ismael De la Rosa-Vargas. Analysis of a multiclass classification problem by lasso logistic regression and singular value decomposition to identify sound patterns in queenless bee colonies. Computers and Electronics in Agriculture, 159:69–74, 2019.
(c) GdR 720 ISIS - CNRS - 2011-2020.