Selected Research Projects

Computer Vision · 3D Geometry · Neural Pose Estimation

• Evaluating robust neural solvers for camera pose estimation under high outlier ratios and image-noise perturbations.
• Investigating top-K hypothesis aggregation and quaternion rotation averaging to improve pose stability and robustness.

Medical Image Segmentation · U-Net · U-Net++ · Transfer Learning

• Studied retinal vessel segmentation on DRIVE and CHASE datasets using U-Net and U-Net++ architectures with pretrained backbones.
• Evaluated models using Dice, IoU, sensitivity, specificity, and accuracy, with emphasis on robust medical image segmentation.

Cybersecurity · Explainable Machine Learning · Ensemble Learning

• Designed machine learning pipelines for phishing website detection using ensemble-based classification strategies.

Natural Language Processing · Social Media Mining · Deep Learning

• Built sentiment analysis models to study public opinion during the COVID-19 pandemic using Twitter data.