Attention Refined Multi-Scale Enhanced CNN For Accurate and Lightweight Leaf Disease Detection in Smart Precision Agriculture
- Version
- Download 24
- File Size 505.50 KB
- File Count 1
- Create Date 10 March 2026
- Last Updated 10 March 2026
Attention Refined Multi-Scale Enhanced CNN For Accurate and Lightweight Leaf Disease Detection in Smart Precision Agriculture
Dr. S. Joy Kumar1, Mrs. Mulakalapally Shailaja2, Dr.Kancharla Bullibabu3,
Mrs. Masi reddy Sadalaxmi4
1Department of CSE, St. Mary's Engineering College, Hyderabad, India. (Orcid ID: https://orcid.org/0009-0007-8494-4269 )
2Department of CS, MJPTBCWRDC(W), Suryapet, Telangana, India. (Orcid ID: https://orcid.org/0009-0001-4990-4946 )
3Department of CSE, Anurag Engineering College, Kodad, Suryapet, India. (Orcid ID: https://orcid.org/0000-0001-5471-0225 )
4Department of CSE, Anurag Engineering College, Kodad, Suryapet, India. (Orcid ID: https://orcid.org/0009-0007-2853-8178 )
Email address: joykumar@stmarysgroup.com, m.shailaja005@gmail.com , phinehas310@gmail.com, sadalaxmi.morthala@gmail.com
Abstract: This paper proposes an advanced convolutional neural network (ECNN) for the automatic detection and classification of leaf diseases towards precision agriculture. The proposed approach was tested on a standard plant leaf dataset with 54,306 images belonging to 38 classes of leaf diseases, with an 80:20 train/test split. Multi-scale convolutional layers and an attention refinement layer were incorporated to enhance the robustness of the approach in complex field environments. The experimental results show that the proposed ECNN approach resulted in an overall classification accuracy of 98.7%, with precision, recall, and F1-measure values of 98.4%, 98.6%, and 98.5%, respectively. Compared with the latest state-of-the-art models such as EnConv, EfficientNet-B0, and hybrid CNN models, the proposed approach improved the accuracy by 2.1% and reduced the computational complexity by 18.3%. The proposed approach was also found to be reliable in early-stage disease detection and under varying illumination conditions, making it suitable for real-time implementation in smart farming systems.
Keywords: Leaf Disease Detection, Enhanced CNN, Precision Agriculture, Attention Mechanism and Deep Learning
Download