Update README.md
Browse files
README.md
CHANGED
|
@@ -11,13 +11,22 @@ tags:
|
|
| 11 |
---
|
| 12 |
|
| 13 |
# AIGVDet: AI-Generated Video Detection via Spatial-Temporal Anomaly Learning
|
|
|
|
| 14 |
## Dataset Summary
|
|
|
|
| 15 |
This dataset is the official data repository for the paper **"AI-Generated Video Detection via Spatial-Temporal Anomaly Learning"**, accepted at the 7th Chinese Conference on Pattern Recognition and Computer Vision (PRCV 2024).
|
|
|
|
| 16 |
It contains the preprocessed training frames and test videos designed for training and evaluating models that detect AI-generated videos by learning spatial-temporal anomalies. The method utilizes both Spatial Domain Detectors (working on RGB frames) and Optical Flow Detectors (working on optical flow frames) to distinguish between real and fake (AI-generated) videos.
|
|
|
|
| 17 |
### Original Source Code
|
| 18 |
The official implementation code and further usage instructions can be found in the [GitHub Repository: multimediaFor/AIGVDet](https://github.com/multimediaFor/AIGVDet).
|
|
|
|
| 19 |
## Dataset Structure
|
| 20 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
```text
|
| 22 |
data/
|
| 23 |
├── train/
|
|
@@ -30,20 +39,25 @@ data/
|
|
| 30 |
│ ├── video_00000/
|
| 31 |
│ │ ├── 00000.png
|
| 32 |
│ │ └── ...
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
│ ├── 0_real/
|
| 36 |
-
│ └── 1_fake/
|
| 37 |
-
└── test/
|
| 38 |
-
└── testset_1/
|
| 39 |
├── 0_real/
|
| 40 |
└── 1_fake/
|
| 41 |
```
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
## Intended Use & License
|
|
|
|
| 43 |
The code and dataset are released **only for academic research**. Commercial usage is **strictly prohibited**.
|
|
|
|
| 44 |
If you wish to use this dataset, please adhere to these restrictions and respect the original authors' licensing terms.
|
|
|
|
| 45 |
## Citation
|
|
|
|
| 46 |
If you find this dataset or the related code useful in your research, please consider citing the original paper:
|
|
|
|
| 47 |
```bibtex
|
| 48 |
@article{AIGVDet24,
|
| 49 |
author = {Jianfa Bai and Man Lin and Gang Cao and Zijie Lou},
|
|
@@ -52,5 +66,7 @@ If you find this dataset or the related code useful in your research, please con
|
|
| 52 |
year = {2024},
|
| 53 |
}
|
| 54 |
```
|
|
|
|
| 55 |
## Contact
|
|
|
|
| 56 |
For any questions regarding the original paper, code, or dataset, please contact the authors at [lyan924@cuc.edu.cn](mailto:lyan924@cuc.edu.cn).
|
|
|
|
| 11 |
---
|
| 12 |
|
| 13 |
# AIGVDet: AI-Generated Video Detection via Spatial-Temporal Anomaly Learning
|
| 14 |
+
|
| 15 |
## Dataset Summary
|
| 16 |
+
|
| 17 |
This dataset is the official data repository for the paper **"AI-Generated Video Detection via Spatial-Temporal Anomaly Learning"**, accepted at the 7th Chinese Conference on Pattern Recognition and Computer Vision (PRCV 2024).
|
| 18 |
+
|
| 19 |
It contains the preprocessed training frames and test videos designed for training and evaluating models that detect AI-generated videos by learning spatial-temporal anomalies. The method utilizes both Spatial Domain Detectors (working on RGB frames) and Optical Flow Detectors (working on optical flow frames) to distinguish between real and fake (AI-generated) videos.
|
| 20 |
+
|
| 21 |
### Original Source Code
|
| 22 |
The official implementation code and further usage instructions can be found in the [GitHub Repository: multimediaFor/AIGVDet](https://github.com/multimediaFor/AIGVDet).
|
| 23 |
+
|
| 24 |
## Dataset Structure
|
| 25 |
+
|
| 26 |
+
The Hugging Face dataset repository contains the `train` and `val` splits. The `test` set is hosted separately.
|
| 27 |
+
|
| 28 |
+
Within each split, the data is further categorized into `0_real` and `1_fake` classes. Each video is represented by a folder containing its extracted frames.
|
| 29 |
+
|
| 30 |
```text
|
| 31 |
data/
|
| 32 |
├── train/
|
|
|
|
| 39 |
│ ├── video_00000/
|
| 40 |
│ │ ├── 00000.png
|
| 41 |
│ │ └── ...
|
| 42 |
+
└── val/
|
| 43 |
+
└── val_set_1/
|
|
|
|
|
|
|
|
|
|
|
|
|
| 44 |
├── 0_real/
|
| 45 |
└── 1_fake/
|
| 46 |
```
|
| 47 |
+
|
| 48 |
+
### Test Set
|
| 49 |
+
The test videos can be downloaded from [Google Drive](https://drive.google.com/drive/folders/1D84SRWEJ8BK8KBpTMuGi3BUM80mW_dKb?usp=sharing).
|
| 50 |
+
|
| 51 |
## Intended Use & License
|
| 52 |
+
|
| 53 |
The code and dataset are released **only for academic research**. Commercial usage is **strictly prohibited**.
|
| 54 |
+
|
| 55 |
If you wish to use this dataset, please adhere to these restrictions and respect the original authors' licensing terms.
|
| 56 |
+
|
| 57 |
## Citation
|
| 58 |
+
|
| 59 |
If you find this dataset or the related code useful in your research, please consider citing the original paper:
|
| 60 |
+
|
| 61 |
```bibtex
|
| 62 |
@article{AIGVDet24,
|
| 63 |
author = {Jianfa Bai and Man Lin and Gang Cao and Zijie Lou},
|
|
|
|
| 66 |
year = {2024},
|
| 67 |
}
|
| 68 |
```
|
| 69 |
+
|
| 70 |
## Contact
|
| 71 |
+
|
| 72 |
For any questions regarding the original paper, code, or dataset, please contact the authors at [lyan924@cuc.edu.cn](mailto:lyan924@cuc.edu.cn).
|