Improving Fine-Grained Rice Leaf Disease Detection via Angular-Compactness Dual Loss Learning

Md. Rokon Mia; Rakib Hossain Sajib; Abdullah Al Noman; Abir Ahmed; B M Taslimul Haque

cs.CVMar 26, 2026

Improving Fine-Grained Rice Leaf Disease Detection via Angular-Compactness Dual Loss Learning

Authors: Md. Rokon Mia, Rakib Hossain Sajib, Abdullah Al Noman, Abir Ahmed, B M Taslimul Haque

Organizations: 1,2Department of Computer Science and Engineering, Begum Rokeya University, Rangpur, Rangpur, Bangladesh · 3Wilmington University, New Castle, Delaware, United States · 4Washington University of Science and Technology, Alexandria, Virginia, United States · 5Central Michigan University, Mount Pleasant, Michigan, United States

Abstract

Early detection of rice leaf diseases is critical, as rice is a staple crop supporting a substantial share of the world's population. Timely identification of these diseases enables more effective intervention and significantly reduces the risk of large-scale crop losses. However, traditional deep learning models primarily rely on cross entropy loss, which often struggles with high intra-class variance and inter-class similarity, common challenges in plant pathology datasets. To tackle this, we propose a dual-loss framework that combines Center Loss and ArcFace Loss to enhance fine-grained classification of rice leaf diseases. The method is applied into three state-of-the-art backbone architectures: InceptionNetV3, DenseNet201, and EfficientNetB0 trained on the public Rice Leaf Dataset. Our approach achieves significant performance gains, with accuracies of 99.6%, 99.2% and 99.2% respectively. The results demonstrate that angular margin-based and center-based constraints substantially boost the discriminative strength of feature embeddings. In particular, the framework does not require major architectural modifications, making it efficient and practical for real-world deployment in farming environments.

Figures & tables

Fig. 1: Representative samples of rice leaf disease classes.

Fig. 2: Schematic representation of the proposed framework for plant disease classification.

Category	Train	Validation	Test
Bacterial Leaf Blight	439	73	35
Brown Spot	522	75	46
Healthy	557	100	47
Leaf Blast	522	77	51
Leaf Scald	412	88	35
Narrow Brown Spot	521	76	36

TABLE I: Dataset Image Distribution by Category and Split

\mathbf{F}_{i}=\mathrm{Backbone}(\mathbf{X}_{i}),\quad\mathbf{F}_{i}\in\mathbb{R}^{m\times C\times H^{\prime}\times W^{\prime}}

Algorithm 1 Training Procedure with Dual-Loss Supervision

Parameter	Value
Optimizer	AdamW (learning rate = 0.0001)
Loss Functions	ArcFace, Center Loss
Batch Size	32
Epochs	30
Regularization	Dropout (0.5), L2 regularization ( $\lambda=0.0001$ )
Scale Factor ( $s$ )	30

TABLE II: Training Configuration

Authour	Year	Dataset	Method	Accuracy (%)
Catal Reise and Turk et al. [ 20 ]	2024	Wheat Leaf Dataset	RegNetY080	97.64
Abasi et al. [ 17 ]	2023	Rice Leaf Images	Custom CNN	91.40
Ahad et al. [ 21 ]	2023	Rice Leaf Dataset (UCI)	DEX (Densenet121, EfficientNetB7, and Xception)	98.00
Sankareshwaran et al. [ 22 ]	2023	Rice Leaf Dataset	CAHA-AXRNet	98.10
Naresh kumar et al. [ 3 ]	2025	Rice Leaf Dataset	FVBC model	97.6
Proposed framework	-	Rice Leaf Dataset	InceptionNetV3+Dual Loss	99.6

TABLE III: Comparison of the proposed method with state-of-the-art models on rice leaf disease classification.

Model	Loss Function	Precision (%)	Recall (%)	F1 Score (%)	Accuracy (%)	Experiment
InceptionNetV3	Cross Entropy	99.31	99.31	99.31	99.20	a
InceptionNetV3	ArcFace + Center Loss	99.65	99.67	99.66	99.60	b
DenseNet201	Cross Entropy	99.29	98.33	99.28	98.00	c
DenseNet201	ArcFace + Center Loss	99.31	99.31	99.31	99.20	d
EfficientNetB0	Cross Entropy	98.19	98.13	98.14	98.00	e
EfficientNetB0	ArcFace + Center Loss	99.31	99.35	99.31	99.20	f

TABLE IV: Performance of pretrained models on Roboflow RLD using Cross Entropy vs. ArcFace + Center Loss.

Fig. 3: Performance comparison of pretrained models with Cross Entropy vs. ArcFace + Center Loss.

No.	$\mathcal{L}_{\mathrm{CE}}$	$\mathcal{L}_{\mathrm{Center}}$	$\mathcal{L}_{\mathrm{Arc}}$	Precision	Recall	F1-Score	Accuracy
1	✓			99.31	99.31	99.31	99.20
2		✓		19.50	38.56	25.07	38.80
3			✓	98.68	98.59	98.62	98.40
4		✓	✓	99.65	99.67	99.66	99.60

TABLE V: Performance of ablation experiment on Roboflow RLD dataset.

No.	$\mathcal{L}_{\mathrm{CE}}$	$\mathcal{L}_{\mathrm{Center}}$	$\mathcal{L}_{\mathrm{Arc}}$	Precision	Recall	F1-Score	Accuracy
1	✓			97.39	97.41	97.38	97.35
2		✓		4.01	16.3	6.43	16.67
3			✓	95.37	95.08	95.16	95.08
4		✓	✓	98.17	98.15	98.13	98.11

TABLE VI: Performance of ablation experiment on Kaggle RLD dataset.

Plant Disease Classification

Plant disease classification uses image and sensor data to identify crop diseases and support timely, targeted agricultural intervention.

See all 62 papers

Cross-Architectural Mixture-of-Experts with Adaptive Soft Routing for Plant Leaf Disease Classification
Jun 22, 2026Phi-Hung Hoang, Thi-Thu-Hong PhanPlant Disease Classification Mixture-of-Experts Models
AIT laboratory, Faculty of Artificial Intelligence, FPT University, Da Nang, 550000, Viet Nam
CottonLeafVision: An Explainable and Robust Deep Learning Framework for Cotton Leaf Disease Classification
Jun 12, 2026Rafi Ahamed, Md. Abir Rahman, Tasnia Tarannum Roza +3Plant Disease Classification Explainable Image Classification
Dept. of CSE East West University Dhaka, Bangladesh
Enhancing Precision Agriculture with a Hybrid Deep Learning Framework for Multi-Class Plant Disease Classification and Interpretability
Jun 13, 2026Hasibul Islam Sufi, Ridam Roy, Shayla Alam Setu +1Plant Disease Classification Hybrid CNN-Transformer Architectures
Department of Computer Science and Engineering, Daffodil International University, Dhaka 1216, Bangladesh

Jun 22, 2026J/K move · Enter open · S save

Cross-Architectural Mixture-of-Experts with Adaptive Soft Routing for Plant Leaf Disease Classification

Phi-Hung Hoang, Thi-Thu-Hong Phan

AIT laboratory, Faculty of Artificial Intelligence, FPT University, Da Nang, 550000, Viet Nam

Plant Disease Classification Mixture-of-Experts Models

Open paper AskPDF arXiv

Plant leaf disease classification is crucial for crop protection and precision agriculture but remains challenging under complex backgrounds, illumination variations, and severe class imbalance. Moreover, single-architecture models often fail to effectively capture both local and global representations. To address these challenges, this study proposes an adaptive soft Mixture-of-Experts (MoE) framework with cross-architectural routing that integrates EfficientNet-B0, DenseNet-121, and Swin-Tiny to exploit complementary multi-scale, local, and global features. A soft gating mechanism dynamically assigns input-dependent expert weights, while a two-stage refinement training strategy improves optimization stability and generalization. Experiments on a highly imbalanced potato leaf disease dataset achieve 91.68% recall and 92.62% F1-score, surpassing the strongest individual expert by 5.91% and 5.03%, respectively. Additional evaluations on durian and sesame leaf disease datasets yield F1-scores of 94.03% and 97.04%, demonstrating robust cross-dataset generalization and the potential of the proposed framework for reliable real-world crop health monitoring

Abstract

Plant Disease Classification

Explore similar work

Cross-Architectural Mixture-of-Experts with Adaptive Soft Routing for Plant Leaf Disease Classification

CottonLeafVision: An Explainable and Robust Deep Learning Framework for Cotton Leaf Disease Classification

Enhancing Precision Agriculture with a Hybrid Deep Learning Framework for Multi-Class Plant Disease Classification and Interpretability