Sound Classification using Convolutional Neural Network: A Prototyping Tool for Ganoderma Oil Palm Disease Detection

• Item type: Symposium(1)
• 公開日: 2022-01-28
• タイトル: Sound Classification using Convolutional Neural Network: A Prototyping Tool for Ganoderma Oil Palm Disease Detection
• タイトル / 言語: en
• タイトル: Sound Classification using Convolutional Neural Network: A Prototyping Tool for Ganoderma Oil Palm Disease Detection
• 言語: eng
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_5794
• 資源タイプ: conference paper
• 著者所属: Department of Mechanical and Mechatronics Engineering, Prince of Songkla University
• 著者所属: Department of Mechanical and Mechatronics Engineering, Prince of Songkla University
• 著者所属: Department Electrical Engineering , Faculty of Engineering Prince of Songkla University
• 著者所属(英): en
• 著者所属(英): Department of Mechanical and Mechatronics Engineering, Prince of Songkla University
• 著者所属(英): en
• 著者所属(英): Department of Mechanical and Mechatronics Engineering, Prince of Songkla University
• 著者所属(英): en
• 著者所属(英): Department Electrical Engineering , Faculty of Engineering Prince of Songkla University
• 著者名: Chalatorn, Augsornthip; Paramin, Neranon; Pornchai, Phukpattaranont
• 著者名(英): Chalatorn, Augsornthip; Paramin, Neranon; Pornchai, Phukpattaranont
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: Oil palm trees are one of the most important crops in Southeast Asia. Palm oil is capable of creating products that are important for the economy and citizens' consumption. One of the most major problems of oil palm diseases is Ganoderma disease, which is a white-rot fungus causing oil palm trees to wilt and die. Conventionally, local wisdom is initially executed to classify the oil palm tree disease using the sound technique by knocking the trunk. This is because the Ganoderma disease can massively damage a stem of the oil palm tree. Consequently, this research is aiming to develop a prototype device to be able to initially diagnose the oil palm disease. The results show the optimized experimental setup for the sound recording process i.e. the knocking force acting to the tree trunk, the position of the microphone and the trunk position of the impact point. Finally, the Convolutional Neural Network (CNN) was successfully implemented for classifying the wood-knocking sounds based on the generated spectrograms of these sounds with an achieved accuracy of approximately 70-80%. It can be concluded that the proposed method based on the CNN approach in the sound classification and recognition systems can be used to initially diagnose the Ganoderma disease.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Oil palm trees are one of the most important crops in Southeast Asia. Palm oil is capable of creating products that are important for the economy and citizens' consumption. One of the most major problems of oil palm diseases is Ganoderma disease, which is a white-rot fungus causing oil palm trees to wilt and die. Conventionally, local wisdom is initially executed to classify the oil palm tree disease using the sound technique by knocking the trunk. This is because the Ganoderma disease can massively damage a stem of the oil palm tree. Consequently, this research is aiming to develop a prototype device to be able to initially diagnose the oil palm disease. The results show the optimized experimental setup for the sound recording process i.e. the knocking force acting to the tree trunk, the position of the microphone and the trunk position of the impact point. Finally, the Convolutional Neural Network (CNN) was successfully implemented for classifying the wood-knocking sounds based on the generated spectrograms of these sounds with an achieved accuracy of approximately 70-80%. It can be concluded that the proposed method based on the CNN approach in the sound classification and recognition systems can be used to initially diagnose the Ganoderma disease.
• 書誌情報: Proceedings of Asia Pacific Conference on Robot IoT System Development and Platform; 巻 2021, p. 88-89, 発行日 2022-01-28
• 出版者 / 言語: ja
• 出版者: 情報処理学会