Model Card for

NTIRE2026-Robust Deepfake Detection Challenge - Ant International

model_weight:

  • ViT-CLS (Weight file: deepfake_v135_epoch_1.pt)
  • ViT-Attnpool (Weight file: deepfake_v156_epoch_1.pt)

How to Use

1. Prerequisites

  • Python 3.10 is recommended for compatibility.
  • Ensure all dependencies from requirements.txt are installed.

2. Configure Inference

  • Open the infer.sh script in a text editor.
  • Locate the -d flag and change its argument to the full path of your test image directory.

3. Run and Get Results

  • Execute the script from your terminal: bash infer.sh.
  • Upon completion, a submission.zip file containing the predictions will be created in the root directory.
Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for scarlettss/NTIRE2026_deepfake

Dataset used to train scarlettss/NTIRE2026_deepfake