DEVELOPMENT OF NON-DESTRUCTIVE ROBOT FOR EGGPLANTS (SOLANUM MELONGENA) PEST DETECTION AND CLASSIFICATION
Effective pest monitoring is essential for farmers to detect potential crop damage, minimize the use of pesticides and insecticides, and optimize harvest yields. Deep learning techniques, such as convolutional neural networks, enable accurate pest identification and decision-making based on insect i...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
University of Kragujevac
2025-06-01
|
| Series: | Proceedings on Engineering Sciences |
| Subjects: | |
| Online Access: | https://pesjournal.net/journal/v7-n2/64.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Effective pest monitoring is essential for farmers to detect potential crop damage, minimize the use of pesticides and insecticides, and optimize harvest yields. Deep learning techniques, such as convolutional neural networks, enable accurate pest identification and decision-making based on insect images. Automation through artificial intelligence (AI) provides opportunities to automate labor-intensive insect control techniques, improving efficiency and reducing risks associated with manual labor. This research proposes a real-time, scalable pest monitoring system that leverages computer vision and machine learning for field crop insect identification. To achieve this goal, a camera-based monitoring system was developed that captures images of plants at regular intervals, analyzes them using software, and categorizes the pests on the images. The camera is mounted on a self-moving robot with a self-powering mechanism using a solar panel attached to a portable battery. The results of pest classification are sent to a mobile application via Firebase. The pest classification model developed yielded notable recall rates, notably 76% for caterpillars, 93% for flea beetles, and 89% for other pests. Additionally, with an average prediction test accuracy of 86%, the researchers have deemed the model's performance relatively satisfactory. However, despite these promising findings, the researchers have outlined various recommendations for enhancing future studies in this field. |
|---|---|
| ISSN: | 2620-2832 2683-4111 |