Aesthetic Evaluation Framework

This repository is intended to aggregate several aesthetic evaluation methods into a simple framework.

Each model's interface provides predict_from_pil and predict_from_tensor methods, allowing both PIL and tensor formats for the input image. The tensor format is particularly useful when gradient calculation is required.

Run example.py for testing the framework.

Setup

1. Clone the repository including submodules:

   If you haven't cloned the repository yet, run:

   git clone --recursive <repository-url>

   If you've already cloned the repository without submodules, run:

   git submodule update --init --recursive

2. Create the conda environment:

   conda env create -f environment.yml

3. Install openclip: python -m pip install open_clip_torch

Current Models

• LAION Aesthetic Predictor v1 and v2
• Simulacra Aesthetic Predictor

Usage Example

python example.py <image_path>