VADB: Video Aesthetics Database and Scoring Framework

This repository introduces VADB, a large-scale video aesthetics database, and VADB-Net, a novel video aesthetics scoring framework designed to evaluate the aesthetic quality of videos across multiple dimensions. Our work provides comprehensive resources for researchers and developers interested in video aesthetics analysis, computer vision, and multimedia content assessment.

Figure 1: Architecture of the VADB-Net Video Encoder

Figure 2: Sample videos and annotations from the VADB dataset

📜 License

• VADB dataset: Licensed under CC BY-NC 4.0 (commercial use prohibited).
• Code and models: Licensed under CC BY 4.0

📦 Dataset

The VADB dataset is publicly available on Hugging Face:
https://huggingface.co/datasets/BestiVictoryLab/VADB

The VADB dataset is licensed under CC BY-NC 4.0, which prohibits any commercial use. For commercial collaborations, please contact us.

It includes:

• 7,881 videos covering diverse video styles and content categories
• Detailed language comments for each video
• Aesthetic scores across 7-11 dimensions, comprehensively covering the aesthetic attribute features of videos
• Rich objective tags, annotating video shooting techniques and other objective dimensions

🧠 Models & Code

Video Encoder

The pre-trained video encoder model can be obtained from Google Drive:
drive_link

This encoder extracts aesthetic feature vectors from videos and serves as the foundational component for all scoring models.

Scoring Models

The repository is structured into three main components:

1. Overall Aesthetic Score

• Folder: 1TotalScore
• Model: Predicts the overall aesthetic score of videos
• Usage:

  cd 1TotalScore
  python 1TotalScore.py

2. General Attribute Scores

• Folder: 2GeneralAttribute
• Model: Evaluates general aesthetic attributes of videos
• Evaluation Dimensions:
  • Composition
  • Shot Size
  • Lighting
  • Visual Tone
  • Color
  • Depth of Field
• Usage:

  cd 2GeneralAttribute
  python 2GeneralAttribute.py

3. Human-Centric Attribute Scores

• Folder: 3HumanAttribute
• Model: Focuses on evaluating specific aesthetic attributes of human subjects
• Evaluation Dimensions:
  • Expression
  • Movement
  • Costume
  • Makeup
• Usage:

  cd 3HumanAttribute
  python 3HumanAttribute.py

🚀 Getting Started

1. Install the required dependencies (From CLIP):

   conda install --yes -c pytorch pytorch=1.7.1 torchvision cudatoolkit=11.0  
   pip install ftfy regex tqdm  
   pip install opencv-python boto3 requests pandas  

2. Download the VADB dataset from Hugging Face:

   git lfs install  
   git clone https://huggingface.co/datasets/BestiVictoryLab/VADB  

3. Load the pre-trained video encoder (see Video Encoder).
4. Run the scoring model suitable for your use case (see Scoring Models).