Rice leaf disease classification using resource-efficient deep learning models via response-based knowledge distillation

Maimunul Karim Jisan

doi:10.1016/j.asoc.2026.115356

Rice leaf disease classification using resource-efficient deep learning models via response-based knowledge distillation

2026-04-29

0

PRE

AI

M

Maimunul Karim Jisan *

DOI：10.1016/j.asoc.2026.115356

Original

Original request for help

Save

Abstract

En 中文

• High-parameter CNN is better in distilling knowledge to the student in Rice Leaf Disease Classification. • Respond-Based Knowledge Distillation (RKD) significantly improved the student models misclassification errors. • CNN student trained with RKD using the teachers perform 11% better than without RKD. • VGG16 produced the best-performing CNN student model through RKD. • In RLDIS, the LiteCNN4Rice achieves 98.15% accuracy with only 112kb H5 file size.

Keywords:

Rice leaf disease classification

Response-Based Knowledge Distillation

CNN student models

VGG16

LiteCNN4Rice

Rice leaf disease classification using resource-efficient deep learning models via response-based knowledge distillation

Abstract

Journal

Organization