My research interests are broadly in the areas of visual information processing, deep learning, and computer vision. My current research focus is on visual quality assessment, which develops methods that predict the perceptual quality of an image or video. These methods can be used further in downstream tasks that optimize visual content delivery without compromising the perceptual quality, for example, image/video compression, streaming, enhancement, retrieval, etc.
I develop methods to assess visual quality by employing a variety of tools:
- Deep feature learning
- Self-supervised representation learning (SSL)
- Contrastive vision-language (VL) models
- Multi-modal fusion models
- Multi-modal large language models (MLLM)