Unified Image Quality Assessment for Cross-Content Applications
PhD research at Sunway University on perceptual image quality assessment, including permutation-entropy-based features and cross-content quality modelling.
Selected research and applied projects from my academic and professional experience.
PhD research at Sunway University on perceptual image quality assessment, including permutation-entropy-based features and cross-content quality modelling.
MS research on underwater acoustic signal detection and classification, supported through a Pakistan Navy-funded research project.
Principal Investigator for a project on non-invasive disease diagnosis using digital infrared thermal imaging, funded by ICT R&D, Ministry of Information and Technology, Government of Pakistan.
Industry work involving ResNet and YOLO-based models, object detection, segmentation, classification, hyperparameter optimisation, synthetic augmentation, preprocessing automation and MLOps/data-engineering workflows.
Deep residual network classifier using a modified ResNet-18 with transfer learning and class weighting for natural-scene and screen-content image categories drawn from benchmark IQA databases.
ResNet-50-based regression framework for estimating additive white Gaussian noise variance, using augmentation, dropout and a self-generated image dataset.
Speech feature pipeline using PLP, BFCC, MFCC and GFCC representations with CNN and multilayer feed-forward neural-network classifiers on a custom speaker dataset.
Iris-recognition work using iris-pupil boundary extraction, filtering and morphological operations, with neural-network classifiers evaluated on the CASIA iris database.