Coffee Leaf Disease Recognition Based on Deep Learning and Texture Attributes
| Autor: | Roseli dos Reis Goulart, Francisco Fambrini, Jose Hiroki Saito, Lucas Ximenes Boa Sorte, Carolina Toledo Ferraz |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
Local binary patterns
Computer science business.industry Deep learning 020206 networking & telecommunications Pattern recognition 02 engineering and technology Texture (music) computer.software_genre Convolutional neural network Expert system Plant disease Pattern recognition (psychology) 0202 electrical engineering electronic engineering information engineering General Earth and Planetary Sciences Feedforward neural network 020201 artificial intelligence & image processing Artificial intelligence business computer General Environmental Science |
| Zdroj: | KES |
| ISSN: | 1877-0509 |
| DOI: | 10.1016/j.procs.2019.09.168 |
| Popis: | An automatic coffee plant disease recognition system is required since coffee is an important commodity in the world economy and its productivity and quality are affected by diseases such as Cercospora and Rust. This research aims to apply computational methods to recognize main diseases in coffee leaves, with the purpose to implement an expert system to assist coffee producers in disease diagnosis during its initial stages. Since these two diseases are shapeless, it inspires a texture attribute extraction approach for pattern recognition. Two texture attributes were considered in this work: statistical attributes and local binary patterns. The texture attribute vector were computed for a collection of images of coffee leaves and used as input to a feedforward neural network. The results were compared with the recognition rate of a convolutional neural network with deep learning applied directly to the same collection of images, without extraction of texture attributes. Surprisingly, this second approach showed better results than the texture extraction method. It could be explained by the small number of diseases we aimed to recognize and a sufficient number of training samples used during the deep learning process. The best Kappa coefficient obtained was 0.970, and sensitivity was 0.980. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|