egorchistov/sr-artifact-detection-wasd

WASD: Weakly-supervised Artifact Scoring with Discriminative triplet loss

💻 Code | 🤗 Demo

🔍 Find where super-resolution goes wrong and which artifacts appear

🛠️ Usage

Install srflawscry via the package manager:

pip3 install git+https://github.com/msu-video-group/srflawscry

Then use the following snippet to run WASD metric on super-resolution method results:

from pathlib import Path

from srflawscry import WASD

model = WASD.from_pretrained("egorchistov/sr-artifact-detection-wasd").eval()
model.run(
    image_path=Path("result.png"),
    source_path=Path("source.png"),
    output_path=Path("artifact_mask.png"),
)