DeepFake Detection
DeepFake Detection is the task of detecting fake videos or images that have been generated using deep learning techniques. Deepfakes are created by using machine learning algorithms to manipulate or replace parts of an original video or image, such as the face of a person. The goal of deepfake detection is to identify such manipulations and distinguish them from real videos or images.
Description source: DeepFakes: a New Threat to Face Recognition? Assessment and Detection
Image source: DeepFakes: a New Threat to Face Recognition? Assessment and Detection
Papers
No papers found.
All datasetsFakeAVCelebFaceForensicsDFDC1CIFAKE: Real and AI-Generated Synthetic ImagesCOCOFakeDFFDLAV-DF
Benchmark Results
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | AV-Lip-Sync+ | Accuracy (%) | 99.29 | — | Unverified |
| 2 | Avtenet | Accuracy (%) | 98.57 | — | Unverified |
| 3 | FACTOR | ROC AUC | 97.4 | — | Unverified |
| 4 | RealForensics | ROC AUC | 97.1 | — | Unverified |
| 5 | AVAD | ROC AUC | 94.5 | — | Unverified |
| 6 | AV-Lip-Sync Model | Accuracy (%) | 94 | — | Unverified |
| 7 | FTCN | ROC AUC | 93.1 | — | Unverified |
| 8 | LipForensics | ROC AUC | 91.1 | — | Unverified |
| 9 | Multimodal Ensemble Model | Accuracy (%) | 89 | — | Unverified |
| 10 | AD DFD | ROC AUC | 88.1 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | XceptionNet | DF | 96.36 | — | Unverified |
| 2 | QAD-E | AUC | 0.96 | — | Unverified |
| 3 | EfficientNetB4 + EfficientNetB4ST + B4Att + B4AttST | AUC | 0.94 | — | Unverified |
| 4 | MARLIN (ViT-L) | AUC | 0.94 | — | Unverified |
| 5 | MARLIN (ViT-B) | AUC | 0.93 | — | Unverified |
| 6 | MARLIN (ViT-S) | AUC | 0.89 | — | Unverified |
| 7 | EfficientNetB4 + EfficientNetB4ST + B4AttST | LogLoss | 0.33 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | Cross Efficient Vision Transformer | AUC | 0.95 | — | Unverified |
| 2 | Efficient Vision Transformer | AUC | 0.92 | — | Unverified |
| 3 | EfficientNetB4 + EfficientNetB4ST + B4Att | LogLoss | 0.46 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | STYLE | 0L | 99 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | FasterThanLies | AUC | 99.65 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | FasterThanLies | AUC | 1 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | FasterThanLies | AUC | 1 | — | Unverified |
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | BA-TFD | AUC | 0.99 | — | Unverified |