Paddy Leaf Disease Classification Using EfficientNet B4 With Compound Scaling and Swish Activation: A Deep Learning Approach

Abstract

This research presents an advanced paddy disease classification model utilizing the EfficientNet B4 deep learning architecture trained on the publicly available Paddy Doctor dataset, which includes 19,131 labeled images for training and 4,785 images for testing. The model aims to classify paddy leaf samples into nine disease categories or into normal specimens. EfficientNet B4, known for its structured scaling technique and Swish activation function, was trained using pretrained weights from ImageNet and optimized with the Adam optimizer. It achieved a peak accuracy of 99.09% during training and 96.91% during testing, highlighting its superior performance in image classification tasks. The model’s architecture, featuring mobile inverted bottleneck convolution (MBConv) and deep separable convolutions, facilitated efficient feature extraction, resulting in high accuracy, recall, and F1-scores, along with low loss values. Our findings demonstrate the model’s reliability and efficiency in diagnosing various paddy illnesses, underscoring the transformative potential of deep learning techniques such as EfficientNet B4 in enhancing agricultural disease management.

Keywords

• Paddy disease classification
• EfficientNet B4
• Swish activation
• deep learning
• structured scaling