• Vision Language Model for Pathology (REG 2025): [cite_start]Developed a vision-language model to generate automated diagnostic reports from gigapixel whole slide images[cite: 73].
• Prognostic Risk Stratification of Gliomas: [cite_start]Utilized various pathology foundation models and Multiple Instance Learning (MIL) models for risk stratification in adult-type diffuse gliomas[cite: 74, 75].
• Biomarker Identification in Gliomas: [cite_start]Explored Self-Supervised Learning (SSL) algorithms based on Vision Transformers (ViT) and DINO for biomarker prediction using a weakly supervised approach[cite: 80, 81, 82].
• Automatic Number Plate Recognition: [cite_start]Implemented an end-to-end real-time system using YOLO (You Only Look Once) object detection and Mask-RCNN based Region Proposal Networks[cite: 86, 87, 88].
• Retinal Fundus Glaucoma Analysis (REFUGE2): [cite_start]Leveraged EfficientNets and UNet architectures for optic disc/cup segmentation and glaucoma classification[cite: 89, 90].
• Anomaly Pattern Segmentation in Agriculture: [cite_start]Developed a Feature Pyramid Network (FPN) encoder with varying decoder backbones for aerial image pattern recognition[cite: 91, 92].
• Semantic Scene Segmentation (Indian Driving Dataset): [cite_start]Created an Encoder-Decoder architecture using Inception ResNetV2 and EfficientNets for unstructured environment scene segmentation[cite: 94].
• Skin Lesion Analysis (ISIC Challenge): [cite_start]Developed a Generative Adversarial Network (GAN) based architecture for segmentation, paired with ResNets and MobileNets for classification[cite: 95, 96].