Up to one third of the global food production depends on the pollination of honey bees, making them vital. This study defines a methodology to create a bee hive health monitoring system through image processing techniques. The approach consists of two models, where one performs the detection of bees in an image and the other classifies the detected bee's health. The main contribution of the defined methodology is the increased efficacy of the models, whilst maintaining the same efficiency found in the state of the art. Two databases were used to create models based on Convolutional Neural Network (CNN). The best results consist of 95% accuracy for health classification of a bee and 82% accuracy in detecting the presence of bees in an image, higher than those found in the state-of-the-art.
An intelligent monitoring system for assessing bee hive health / D. Braga, A. Madureira, F. Scotti, V. Piuri, A. Abraham. - In: IEEE ACCESS. - ISSN 2169-3536. - 9(2021 Jun 15), pp. 89009-89019. [10.1109/ACCESS.2021.3089538]
An intelligent monitoring system for assessing bee hive health
F. Scotti;V. Piuri
;
2021
Abstract
Up to one third of the global food production depends on the pollination of honey bees, making them vital. This study defines a methodology to create a bee hive health monitoring system through image processing techniques. The approach consists of two models, where one performs the detection of bees in an image and the other classifies the detected bee's health. The main contribution of the defined methodology is the increased efficacy of the models, whilst maintaining the same efficiency found in the state of the art. Two databases were used to create models based on Convolutional Neural Network (CNN). The best results consist of 95% accuracy for health classification of a bee and 82% accuracy in detecting the presence of bees in an image, higher than those found in the state-of-the-art.File | Dimensione | Formato | |
---|---|---|---|
BEE_09455393.pdf
accesso aperto
Tipologia:
Publisher's version/PDF
Dimensione
1.82 MB
Formato
Adobe PDF
|
1.82 MB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.