Reproduce “EyeSim-VQA: A Free-Energy-Guided Eye Simulation Framework for Video Quality Assessment” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhaoyang Wang, Wen Lu, Jie Li, Lihuo He, Maoguo Gong, Xinbo Gao
A Video Quality Assessment (VQA) framework from Xidian University simulates the human visual system's self-repair and gaze mechanisms, integrating free-energy-guided enhancement and dual-branch processing. The EyeSim-VQA model achieves state-of-the-art performance across diverse VQA benchmarks by approximating the human visual experience.
Key insights
- Explicitly modeling the human eye's free-energy-guided self-repair mechanism and sweeping gaze dynamics can lead to more accurate and perceptually aligned Video Quality Assessment.
- Adaptive visual enhancement, even with lightweight networks, can effectively refine degraded video content and improve feature discriminability for quality prediction.
- A dual-branch approach, combining global aesthetic evaluation with fine-grained technical analysis, along with specific mechanisms for gaze (attention) and scanning (sequential exploration), effectively mimics human visual perception.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.