BrightRate: Quality Assessment for User-Generated HDR Videos

📌 Overview

BrightVQ is the a large-scale subjective video quality dataset dedicated to HDR User-Generate-Content (UGC) video quality assessment. It consists of 300 HDR source videos and 2,100 transcoded versions, with 73,794 subjective quality ratings collected through crowd-sourced subjective study. BrightVQ serves as a benchmark for No-Reference (NR) UGC HDR-VQA models.

Based on BrightVQ, we introduce BrightRate, a novel No-Reference (NR) Video Quality Assessment (VQA) model designed to capture UGC-specific distortions and HDR-specific artifacts.

BrightRate integrates UGC-specific features from a pretrained CONTRIQUE model， semantic cues from a CLIP-based encoder,
HDR features extracted via a piecewise non-linear luminance transform, and temporal differences, which are regressed to MOS.

📊 Performance Comparison (SROCC)

✨ Run BrightRate

Please follow --> Demo_Inference

⬇️ Downloading our BrightVQ Dataset

🔗 Downloading the entire dataset

Direct download link for dataset: COMING SOON

🔗 Accessing dataset videos

1️⃣ Directly from AWS S3 (via Browser)

Each video is hosted on AWS S3 and can be accessed using:

wget https://ugchdrmturk.s3.us-east-2.amazonaws.com/videos/VIDEO.mp4 

Replace VIDEO with a hashed video ID from BrightVQ.csv or BrightVQ.txt.

Example:
Police: https://ugchdrmturk.s3.us-east-2.amazonaws.com/videos/ad8affdd94b3c44ae83169fb668ea5c6.mp4

2️⃣ Downloading Videos Using AWS CLI

To download all videos:

cat BrightVQ.txt | while read video; do
    aws s3 cp s3://ugchdrmturk/videos/${video}.mp4 ./BrightVQ/
done

To download a single video:

aws s3 cp s3://ugchdrmturk/videos/VIDEO.mp4 ./BrightVQ/

To download selected videos, create a new text file with list of video IDs:

cat sample-video.txt | while read video; do
    aws s3 cp s3://ugchdrmturk/videos/${video}.mp4 ./BrightVQ/
done

🎬 Sample Videos (Direct Playback)

Below, you can directly play some sample HDR videos from our dataset:

Screen Recording

Train

Dog

Nature

More sample are listed here in table:

Please checkout the full dataset.

🏆 Use Cases and Future Impact

✅ Benchmark for HDR UGC Video Quality Assessment

• BrightVQ provides the first large-scale dataset for HDR UGC quality evaluation, enabling researchers to develop and compare No-Reference (NR) VQA models for HDR conten

✅ Improving HDR Streaming and Compression

• Streaming platforms (e.g., YouTube, Netflix, Prime Video) can optimize their encoding pipelines by using BrightRate and BrightVQ to assess perceptual quality at different bitrates and HDR processing techniques.

✅ Standardization and HDR Quality Metrics Development

• BrightVQ can contribute to the development of new HDR quality standards, potentially influencing ITU-T, MPEG, and industry-led VQA benchmarks

📜 Citation

Please cite us if this work is helpful to you.

COMING SOON

📜 License

BrightVQ is released under a Creative Commons Attribution-NonCommercial (CC BY-NC 4.0) License.

📬 Contact

For questions, please reach out:
📧 [Redacted for Blind Review]