Machine Learning for Agriculture: YOLOv7 for Multiple Tomato Fruit Diseases Detection
Year: 2024
Type: article
Abstract: Tomato fruit diseases are a significant problem for agriculture, causing substantial crop losses globally and impacting productivity. This research explores how YOLOv7 can be utilized to identify and categorize tomato fruit diseases, using a dataset of 2044 labeled images. Advanced image processing techniques such as contrast enhancement and noise reduction were used to train the model and improve detection accuracy. The findings showed an mAP of 89.5%, alongside an 86.2% precision rate and an 82.2% recall rate, indicating the model's strong capability in detecting diseased fruit accurately and reducing false positives. Significantly, the model successfully reached a 97% accuracy rate in identifying Spotted Wilt Virus, a common disease in tomatoes. These results highlight the model’s great capability for practical use in farming settings, where accuracy and timely identification are essential. The research suggests that incorporating machine learning methods like YOLOv7 into disease detection systems can greatly enhance early recognition, decrease crop damage, and lessen the need for chemical treatments. Future efforts will focus on improving the model's ability to identify instances and expanding its use in real-time detection systems in the field.
Source: Nanotechnology Perceptions
Authors: Fernan Mendoza, Thelma Palaoag
Institution: University of the Cordilleras
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
Open Access status: diamond
OpenAlex ID: W4406669538