polito.it
Politecnico di Torino (logo)

Beehive monitoring with computer vision

Cyrian, Douglas, Guy Froissart

Beehive monitoring with computer vision.

Rel. Marcello Chiaberge. Politecnico di Torino, Corso di laurea magistrale in Mechatronic Engineering (Ingegneria Meccatronica), 2023

[img]
Preview
PDF (Tesi_di_laurea) - Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives.

Download (7MB) | Preview
Abstract:

The bee is an essential species for the pollination of many crops; its abnormally high extinction rate is therefore particularly worrying. Domestic bee populations are being carefully observed to identify the reason for this decline. In this context, video hive monitoring aims to enable real-time monitoring while avoiding laborious human labour. We’ll Bee is a start-up seeking to automate certain hive monitoring processes. A tool used to count the number of entries and exits of bees to the hive already exists. The subject of this internship is the improvement of this system in order to determine the proportion of bees bringing pollen back to the hive. Since pollen is used to feed bee larvae, it is indeed a good indicator of the growth of the hive. The observation system consists of a translucent box placed at the entrance to the hive; openings to the outside and to the hive allow the bees to circulate, while a Plexiglas plate ensures that the bees walk at the bottom of this box. A raspberry Pi equipped with a camera attached to the top of the box films the bees at a speed of 15 frames per second. The lighting is natural so as not to disturb the behaviour of the bees. The raspberry is powered by a battery. Ultimately, the objective would be to make the system autonomous by charging the battery with a photovoltaic panel. The objective of this internship was to create a classification algorithm to separate the images of bees bringing pollen to the hive from those not bringing it back. To make such an algorithm useful, video acquisition, image segmentation and tracking of each bee from frame to frame are required. Finally, in order to be able to train an algorithm to classify bees, it is necessary to create a set of catalogued images.

Relatori: Marcello Chiaberge
Anno accademico: 2022/23
Tipo di pubblicazione: Elettronica
Numero di pagine: 83
Soggetti:
Corso di laurea: Corso di laurea magistrale in Mechatronic Engineering (Ingegneria Meccatronica)
Classe di laurea: Nuovo ordinamento > Laurea magistrale > LM-25 - INGEGNERIA DELL'AUTOMAZIONE
Aziende collaboratrici: We'll Bee s.r.l.
URI: http://webthesis.biblio.polito.it/id/eprint/26701
Modifica (riservato agli operatori) Modifica (riservato agli operatori)