←Back to feed
🧠 AI⚪ Neutral
Seeing Beyond 8bits: Subjective and Objective Quality Assessment of HDR-UGC Videos
arXiv – CS AI|Shreshth Saini, Bowen Chen, Neil Birkbeck, Yilin Wang, Balu Adsumilli, Alan C. Bovik||1 views
🤖AI Summary
Researchers introduce Beyond8Bits, a large-scale dataset of 44K HDR user-generated videos with 1.5M crowd ratings, and HDR-Q, the first multimodal large language model designed for HDR video quality assessment. The work addresses limitations of current video quality systems that are optimized for standard dynamic range content.
Key Takeaways
- →Beyond8Bits dataset contains 44K HDR videos from 6.5K sources with over 1.5M crowd quality ratings.
- →HDR-Q is the first multimodal large language model specifically designed for HDR user-generated content video quality assessment.
- →The system uses HDR-aware vision encoders and HDR-Aware Policy Optimization (HAPO) for improved performance.
- →HDR content exposes unique distortions like near-black crushing, highlight clipping, and exposure flicker not handled by SDR models.
- →HDR-Q achieves state-of-the-art performance across HDR video quality assessment benchmarks.
Read Original →via arXiv – CS AI
Act on this with AI
This article mentions $NEAR.
Let your AI agent check your portfolio, get quotes, and propose trades — you review and approve from your device.
Related Articles