http://repo.lib.jfn.ac.lk/ujrr/handle/123456789/128 2026-10-13T12:53:11Z http://repo.lib.jfn.ac.lk/ujrr/handle/123456789/13169 Transformer based Semi-Supervised Learning for Skin Lesion Classification Jathursika, V.; Manivannan, S. Skin cancer is one of the most prevalent cancers worldwide, and early detection is crucial for successful treatment. Deep learning has shown promise in automating skin lesion classification; however, most methods require large volumes of labeled data, which are scarce and costly to obtain in the medical domain. To address this limitation, we propose a Transformer-based Semi- Supervised Learning (SSL) framework for skin lesion classification. Our approach employs the Vision Transformer (ViT) as the backbone model, leveraging its strength in capturing long-range dependencies and global contextual information from dermoscopic images. By integrating SSL, the framework effectively utilizes both labeled and unlabeled data, thereby reducing reliance on extensive manual annotations. We evaluate the proposed method on the publicly available ISIC-2018 dataset. Experimental results show that our SSL framework achieves substantial performance improvements over a fully supervised baseline. Specifically, it achieves an accuracy of 74.3%, balanced accuracy of 62.9%, and F1 score of 62.7representing relative improvements of 9.3%, 11.7%, and 19.9%, respectively, compared to the baseline (65.0% accuracy, 56.3% balanced accuracy, and 52.3% F1 score). These findings demonstrate that SSL can significantly enhance skin lesion classification when labeled data are limited. The proposed Transformerbased SSL framework offers a promising direction for developing cost-effective, reliable, and scalable tools to support early skin cancer detection and improve clinical decision-making. 2025-01-01T00:00:00Z http://repo.lib.jfn.ac.lk/ujrr/handle/123456789/13168 Segmentation of Fruit Surface Defects using Semi Supervised Learning Saarukesan, P.; Manivannan, S. Automated detection and segmentation of fruit-surface defects are essential for improving product quality, reducing dependence on manual inspection, and enabling scalable agricultural automation. Fruit defects vary widely in shape, texture, size, and severity, limiting the generalization of deep learning models. This study evaluates semantic segmentation of orange surface defects using a U-Net encoder–decoder with a ResNet34 backbone, trained on the Orange Navel dataset comprising 1,448 images annotated for three defect categories: sunburn, ulcer, and wind scarring. The model achieves strong quantitative performance: mean Pixel Accuracy (mPA) of 90.57%, mean Intersection over Union (mIoU) of 84.68%, and Dice coefficient of 91.37%. Class-wise IoU highlights near-perfect background segmentation (99.13%) and robust performance on sunburn (85.67%), while ulcer (83.61%) and wind scarring (70.29%) show reduced fidelity-likely due to subtler texture cues, lower contrast, and more irregular lesion boundaries. These results position U-Net with a ResNet34 encoder as a reliable baseline for fruit-defect segmentation. To strengthen generalization under limited annotations, a semi-supervised learning strategy based on the FixMatch paradigm is employed. In this approach, pseudo labels are generated for unlabeled samples, and only high-confidence predictions are used as supervisory signals. Training is guided by enforcing consistency between the model’s predictions and these pseudo labels, allowing effective use of both labeled and unlabeled data and improving segmentation performance across all defect categories. 2025-01-01T00:00:00Z http://repo.lib.jfn.ac.lk/ujrr/handle/123456789/13167 Ordinal-Aware Contrastive Learning for Enhanced Diabetic Retinopathy Grading Suthakaran, A.; Manivannan, S. Diabetic retinopathy (DR) is a leading cause of preventable blindness worldwide. Automated DR grading is inherently challenging due to the ordinal nature of disease stages, where misclassifications across distant stages (e.g., No DR vs. Severe DR) are more critical than between adjacent ones. Existing contrastive and classification-based methods often fail to model these ordinal relationships between severity stages, leading to clinically inconsistent grading outcomes. To address this, we propose an Order-aware Contrastive Loss (OCL) that incorporates inter-class distance weighting to enforce ordinal consistency during representation learning. Furthermore, we introduce a multi-branch architecture built upon a DenseNet-121 backbone, which integrates OCL with two complementary objectives in a unified multi-head framework. The entire network is jointly optimized to enhance both predictive accuracy and ordinal alignment. To the best of our knowledge, this is the first contrastive learning framework that explicitly integrates ordinal distance weighting for diabetic retinopathy grading. The proposed approach is evaluated on two benchmark datasets: EyePACS, a large-scale dataset containing 35,126 training and 42,670 test images across five stages, and APTOS2019, which includes 3,662 images. On EyePACS, our method achieves 86.93% accuracy, 0.8644 quadratic weighted kappa (QWK), and 0.8694 F1- score, surpassing the previous state-of-the-art (84.55% accuracy, 0.8631 QWK, 0.8440 F1-score). Similar improvements are observed on APTOS2019. These results highlight the effectiveness of incorporating order-aware representation learning with multi-objective optimization on a strong DenseNet based architecture, setting a new benchmark for ordinal DR classification. 2025-01-01T00:00:00Z http://repo.lib.jfn.ac.lk/ujrr/handle/123456789/13166 Few-Shot Learning for Early Detection of Coconut Whitefly Infestations using Low-Label Leaf Image Data in Sri Lanka Vigneswaran, V.; Balalojanan, V.; Sukirthan, T.; Thayalan, N.; Thiruthanigesan, K.; Sumathipala, P.; Manivannan, S. Coconut whitefly (Aleurodicus dispersus) has emerged as a devastating pest threatening Sri Lanka’s vital coconut industry, causing severe yield losses and economic hardship for over 1.5 million smallholder farmers. Early detection is critical for effective intervention, but conventional field scouting is labor-intensive and often too late to prevent widespread infestation. While deep learning offers promise for automated pest detection, its deployment is hindered by the scarcity of labeled training data. To address this challenge, we propose a novel SimCLR → FixMatch framework that combines selfsupervised contrastive learning with semi-supervised fine-tuning for accurate detection of coconut whitefly infestations under extreme label scarcity. Our method leverages a new dataset, CocoInfest-SL, comprising 1,600 real-world leaf images from major coconut-growing regions in Sri Lanka, with only 120 labeled samples (30 per class), 7.5% of the full dataset. The model first learns robust feature representations via SimCLR pretraining on both labeled and unlabeled data, then finetunes using FixMatch with confidence-based pseudo-labeling. Experimental results demonstrate that our approach achieves 91.8% accuracy and a 0.91 F1-score, outperforming state-of-theart few-shot and semi-supervised baselines while using minimal annotations. This work presents a scalable, low-cost solution for early pest monitoring in resource-constrained agricultural environments. 2025-01-01T00:00:00Z