Research

This page summarizes my research experience so far.

STR-VQA: Saliency-Driven Transfer Learning with Register Tokens for Spatiotemporal VQA

  • Keywords: Video Quality Assessment, Register Tokens, Saliency Modeling, UNet3D, Spatiotemporal VQA
  • Dr. Mylene Farias, Mayesha Maliha Rahman Mithila
  • February 2025
  • Description: STR-VQA is a no-reference video quality assessment model that combines saliency prediction and register-token-based memory encoding to improve perceptual alignment.

MS-SCANet: A Multiscale Transformer-Based Architecture with Dual Attention for No-Reference Image Quality Assessment

  • Keywords: Image Quality Assessment, Multi-scale attention, Cross-branch attention, Channel attention
  • Dr. Mylene Farias, Mayesha Maliha Rahman Mithila
  • August 2024
  • Citation: DOI: 10.1109/ICASSP49660.2025.10887759

Research Interest

My research interest includes but is not limited to:

  • Image and Video Quality Assessment
  • Visual Attention
  • Deep Learning
  • Computer Vision