| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 286.73 KB | Adobe PDF |
Advisor(s)
Abstract(s)
No âmbito de várias tarefas da investigação apícola, existe uma que obriga o investigador
a classificar e contar o conteúdo de cada favo em cada quadro da colmeia.
Esta tarefa tem por objetivo analisar e controlar a progressão da criação, de abelhas,
e de reservas, o que implica repeti-la múltiplas vezes a cada ano. Cada quadro contém
milhares de favos o que leva a que a contagem, na maior parte dos casos, seja feita de
forma aproximada. Os favos podem conter: pupas (criação fechada), larvas em diferentes
fases de maturação, mel, néctar, pólen, ovos, ou então podem estar vazios. A
automatização deste processo, com o auxílio de um sistema computacional, representa
uma importante evolução na referida tarefa. O presente trabalho aborda a classificação
automática do conteúdo de favos a partir de imagens digitais. Arquiteturas neuronais
de Deep Learning têm mostrado um bom potencial a classificar padrões que exibem
elevada variabilidade visual. Assim, a utilização deste método de aprendizagem máquina
adequa-se à complexidade e variabilidade visual dos padrões apresentados pelas
imagens dos favos. No modelo desenvolvido neste trabalho foi utilizada a arquitetura
neuronal GoogleNet. Esta foi treinada utilizando 63344 imagens anotadas e separadas
nas sete classes referidas. A taxa média de acerto do modelo sobre o conjunto de validação
foi de 94% o que melhora substancialmente o resultado obtido com técnicas
clássicas (SVM - 76%). Este estudo foi financiado pelo projeto BEEHOPE através do concurso
conjunto 2013-2014 BiodivErsA/FACCE-JPI pela FCT (Portugal), CNRS (França) e
MEC (Espanha).
In the apidologie research area, there is one that obliges the researcher to classify and count the contents of each comb cell in each frame. This task aims to analyse and control the progression of brood, bees, and food reserves (honey and pollen), which implies repeating it multiple times a year. Each frame contains thousands of combs which, in most cases, makes the counting inexact and tedious. The combs may contain larvae at different stages, sealed brood, honey, nectar, pollen, eggs or it may be empty. The automation of this process, using a computer system, represents an important evolution in this task. The present work deals with the classification of the contents of comb cells based in digital images. Deep Learning architectures have shown good potential to classify patterns that undergo visual variability. Thus, the use of this machine learning method is adequate to the complexity and visual variability of the patterns presented by the images of the combs. In the developed model, the GoogleNet architecture was used. This was trained using 63,344 images labelled and separated into the seven aforementioned classes. The model hit rate on the validation set was 94% which significantly improved the result obtained using classical techniques (SVM – 76%). This research was funded through the 2013-2014 BiodivERsA/FACCE-JPI Joint call for research proposals, with the national funders FCT (Portugal), CNRS (France), and MEC (Spain).
In the apidologie research area, there is one that obliges the researcher to classify and count the contents of each comb cell in each frame. This task aims to analyse and control the progression of brood, bees, and food reserves (honey and pollen), which implies repeating it multiple times a year. Each frame contains thousands of combs which, in most cases, makes the counting inexact and tedious. The combs may contain larvae at different stages, sealed brood, honey, nectar, pollen, eggs or it may be empty. The automation of this process, using a computer system, represents an important evolution in this task. The present work deals with the classification of the contents of comb cells based in digital images. Deep Learning architectures have shown good potential to classify patterns that undergo visual variability. Thus, the use of this machine learning method is adequate to the complexity and visual variability of the patterns presented by the images of the combs. In the developed model, the GoogleNet architecture was used. This was trained using 63,344 images labelled and separated into the seven aforementioned classes. The model hit rate on the validation set was 94% which significantly improved the result obtained using classical techniques (SVM – 76%). This research was funded through the 2013-2014 BiodivERsA/FACCE-JPI Joint call for research proposals, with the national funders FCT (Portugal), CNRS (France), and MEC (Spain).
Description
Keywords
Apicultura Redes neuronais Processamento de imagens Deep learning
Pedagogical Context
Citation
Alves, Thiago S.; Ventura, Paulo; Neves, Cátia J.; Pinto, M. Alice; Candido Junior, Arnaldo; Paula Filho, Pedro L. De; Rodrigues, Pedro J. (2018). Classificação do conteúdo de favos em quadros de colmeias usando Deep Learning = Classification of honeycomb content in hive frames using deep learning. In V Encontro Jovens Investigadores: livro de resumos. Bragança: Instituto Politécnico. ISBN 978-972-745-235-4
Publisher
Instituto Politécnico de Braganca