Publication: Plant health monitoring using yolov7 object detection
| datacite.subject.fos | oecd::Engineering and technology::Electrical engineering, Electronic engineering, Information engineering::Electrical and electronic engineering | |
| dc.contributor.author | Muhammad Faris Akmal bin Sabri | |
| dc.date.accessioned | 2025-05-20T07:49:01Z | |
| dc.date.available | 2025-05-20T07:49:01Z | |
| dc.date.issued | 2024-08 | |
| dc.description.abstract | This Final Year Project (FYP) titled "Plant Health Monitoring Using YOLOv7 Object Detection" represents a notable advancement in agricultural technology by automating the monitoring of plant health, specifically focusing on eggplant plants. Utilizing the latest object detection algorithms, YOLOv5 and YOLOv7, this project aims to significantly improve the detection of plant health issues. A comprehensive dataset of 4521 images was employed to meticulously train a custom YOLO object detection model to categorize plant leaves as either healthy or stressed. Key quantitative results from the study show that for the "HEALTHY" category, YOLOv7 achieved a precision of 0.897, recall of 0.745, and F1-score of 0.815, compared to YOLOv5, which had a precision of 0.857, recall of 0.667, and F1-score of 0.750. For the "STRESS" category, YOLOv7 demonstrated a precision of 0.611, recall of 0.850, and F1-score of 0.710, whereas YOLOv5 showed a precision of 0.538, recall of 0.700, and F1-score of 0.608 . These results underscore YOLOv7’s superior performance in both categories. The project also highlights the use of transfer learning to enhance the accuracy and training speed of the custom object detector. Transfer learning involves utilizing a pre trained model on a new problem by leveraging knowledge gained from a previous task. This approach helps in efficiently training the model with less data and computational resources while improving its performance. The YOLO object detection process was executed on a laptop GPU connected to a webcam, ensuring seamless and efficient real time monitoring of plant health. This integration of advanced computer vision techniques and deep learning models allows for the timely detection of stress symptoms that are not visible to the naked eye, facilitating prompt interventions and optimizing agricultural practices. The project bridges the gap between traditional manual inspection methods and modern technological solutions, contributing significantly to the advancement of smart farming practices. It offers a reliable tool for proactive plant health management, leading to improved crop yields and sustainable agricultural practices. Future developments aim to create a comprehensive system for monitoring plant health from leaves and analysing environmental conditions, thereby further enhancing the capabilities and applications of this technology in agriculture. | |
| dc.identifier.uri | https://erepo.usm.my/handle/123456789/21750 | |
| dc.language.iso | en | |
| dc.title | Plant health monitoring using yolov7 object detection | |
| dc.type | Resource Types::text::report::technical report | |
| dspace.entity.type | Publication | |
| oairecerif.author.affiliation | Universiti Sains Malaysia |