🔍 Deepfake Detection Space

This space provides an educational, open-source demonstrator collecting a series of baseline state-of-the-art methods for deepfake detection, allowing for free and unlimited use!

If you want to train the detectors in this Space yourself, check out our GitHub repository.

⚠️ Important

  • This tool showcases open-source detectors. It does not reflect the performance, robustness, or accuracy of Truebees' commercial detection methods. Want to give it a try? check Truebees Website
  • These models were trained on image not shared on social networks from the DeepShield dataset. You may experience a drop in accuracy when testing highly compressed images downloaded from social media platforms or messaging apps.
  • You can expect performance comparable to the results shown in Dell'Anna et al. (2025).

⚠️ Due to file size limitations, model weights need to be downloaded automatically on first use. This may take a few moments.
⚠️ To provide a free service, all models run on CPU. The detection process may take a few seconds, depending on the image size and the selected detector.

Individual Detectors

  • Prediction: Tells if an image is Real or Fake.
  • Confidence: The confidence with which the model determines if the image is real or fake.
  • Elapsed Time: The time the model needed to make the prediction (excluding preprocessing or model building).

The "ALL" Option

  • Runs all available detectors (R50_TF, R50_nodown, CLIP-D, P2G, NPR, PGC, ForensicConcept, MDMF) sequentially on the input image.
  • Lists the prediction and confidence of each detector and generates a Confidence Plot visualizing each model's score. No aggregated verdict is produced.
  • In the plot, Green bars indicate a Real prediction, while Red bars indicate a Fake prediction.
Select Detector

Choose which deepfake detection model to use

ALL

  • Description: Runs all available detectors (R50_TF, R50_nodown, CLIP-D, P2G, NPR, PGC, ForensicConcept, MDMF) sequentially on the input image. The complete code for training and testing these models on your own datasets is available in our GitHub repository.
  • Results: Shows the prediction of each detector and a Confidence Plot with each model's score (no aggregated verdict).

R50_TF

R50_nodown

CLIP-D

P2G (Prompt2Guard)

NPR

PGC

  • Description: Peak-guided calibration on a DINOv2-L backbone. Uses the official pre-trained weights released by the authors (not trained on DeepShield).
  • Weights: xiaoyuzhou68/PGC_ckpt

ForensicConcept

  • Description: CLIP ViT-L/14 with a forensic concept codebook. Uses the official pre-trained weights released by the authors (not trained on DeepShield).
  • Weights: ethan225/ForensicConcept · Code: GitHub Repository

MDMF

  • Description: Local distribution shift measured with MMD on DINOv2-L patch embeddings. Uses the official pre-trained weights; the real-image reference bank and decision threshold were built locally from DeepShield data.
  • Code: GitHub Repository

References

  1. Dell'Anna, S., Montibeller, A., & Boato, G. (2025). TrueFake: A Real World Case Dataset of Last Generation Fake Images also Shared on Social Networks. arXiv preprint arXiv:2504.20658.
  2. Corvi, R., et al. (2023). On the detection of synthetic images generated by diffusion models. ICASSP.
  3. Cozzolino, D., et al. (2023). Raising the Bar of AI-generated Image Detection with CLIP. CVPRW.
  4. Laiti, F., et al. (2024). Conditioned Prompt-Optimization for Continual Deepfake Detection. arXiv preprint arXiv:2407.21554.
  5. Tan, C., et al. (2024). Rethinking the up-sampling operations in cnn-based generative network for generalizable deepfake detection. CVPR.