Enhancing Video Quality Models Using Video Understanding (Master)
There has been significant progress in the development of video quality models using DNN-based approaches showing increased accuracy on the datasets considered during the model development and validation process. However, extensive testing of these models reveal drawbacks such as lack of generalizability and explainability. To address this gap, in this thesis, we explore the development of a reasoning-based framework for the development of video quality models. The tasks would include
- Curation of relevant datasets for video quality model development
- Exploration and development of approaches for the development of a reasoning-based framework for video quality modeling
- Performance analysis of the developed model and comparison with state-of-the-art models
Relevant literature (non-exhaustive)
- Cao, L., Sun, W., Zhang, W., Zhu, X., Jia, J., Zhang, K., ... & Min, X. (2026, March). Vqathinker: Exploring generalizable and explainable video quality assessment via reinforcement learning. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 40, No. 4, pp. 2607-2615). https://doi.org/10.48550/arXiv.2508.06051
- Ziheng Jia, Zicheng Zhang, Jiaying Qian, Haoning Wu, Wei Sun, Chunyi Li, Xiaohong Liu, Weisi Lin, Guangtao Zhai, and Xiongkuo Min. 2025. VQA2: Visual Question Answering for Video Quality Assessment. In Proceedings of the 33rd ACM International Conference on Multimedia (MM '25). Association for Computing Machinery, New York, NY, USA, 6751–6760. https://doi.org/10.1145/3746027.3754696
- Mi, Y., Li, Y., Li, Y., Hui, C., Zhang, T., Li, Z., ... & Liu, S. (2025). Q-CLIP: Unleashing the Power of Vision-Language Models for Video Quality Assessment through Unified Cross-Modal Adaptation. arXiv preprint arXiv:2508.06092. https://doi.org/10.48550/arXiv.2508.06092
Suitable for Master theses. Contact persons: Dr.-Ing. Rakesh Rao Ramachandra Rao and Srijeet Roy M.Sc.
