Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
library_name: pytorch
|
| 4 |
+
tags:
|
| 5 |
+
- video-enhancement
|
| 6 |
+
- super-resolution
|
| 7 |
+
- frame-interpolation
|
| 8 |
+
- computer-vision
|
| 9 |
+
pipeline_tag: image-to-image
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Nightingale Video Enhancement Models
|
| 13 |
+
|
| 14 |
+
This repository contains the pre-trained AI models used by the Nightingale Video Enhancement Service.
|
| 15 |
+
|
| 16 |
+
## Models Included
|
| 17 |
+
|
| 18 |
+
### Real-ESRGAN (Super Resolution)
|
| 19 |
+
- **File**: `RealESRGAN_x4plus.pth`
|
| 20 |
+
- **Purpose**: Upscales video resolution (e.g., 480p → 1080p)
|
| 21 |
+
- **Architecture**: Real-ESRGAN x4 plus model
|
| 22 |
+
- **License**: BSD 3-Clause
|
| 23 |
+
- **Source**: [Real-ESRGAN Repository](https://github.com/xinntao/Real-ESRGAN)
|
| 24 |
+
|
| 25 |
+
### RIFE (Frame Interpolation)
|
| 26 |
+
- **File**: `RIFE_v4.6.pth`
|
| 27 |
+
- **Purpose**: Interpolates frames to increase frame rate (e.g., 15fps → 30fps)
|
| 28 |
+
- **Architecture**: RIFE v4.6 model
|
| 29 |
+
- **License**: MIT License
|
| 30 |
+
- **Source**: [RIFE Repository](https://github.com/megvii-research/ECCV2022-RIFE)
|
| 31 |
+
|
| 32 |
+
## Usage
|
| 33 |
+
|
| 34 |
+
These models are automatically downloaded by the Nightingale Video Enhancement Service. To use them in your own project:
|
| 35 |
+
|
| 36 |
+
```python
|
| 37 |
+
from huggingface_hub import hf_hub_download
|
| 38 |
+
|
| 39 |
+
# Download Real-ESRGAN model
|
| 40 |
+
realesrgan_path = hf_hub_download(
|
| 41 |
+
repo_id="maina/nightingale-models",
|
| 42 |
+
filename="RealESRGAN_x4plus.pth"
|
| 43 |
+
)
|
| 44 |
+
|
| 45 |
+
# Download RIFE model
|
| 46 |
+
rife_path = hf_hub_download(
|
| 47 |
+
repo_id="maina/nightingale-models",
|
| 48 |
+
filename="RIFE_v4.6.pth"
|
| 49 |
+
)
|
| 50 |
+
```
|
| 51 |
+
|
| 52 |
+
## Model Details
|
| 53 |
+
|
| 54 |
+
- **Total Size**: ~150MB
|
| 55 |
+
- **Framework**: PyTorch
|
| 56 |
+
- **Device Support**: CUDA GPU (recommended) or CPU
|
| 57 |
+
- **Input**: Video files (MP4, AVI, MOV, etc.)
|
| 58 |
+
- **Output**: Enhanced video with higher resolution and/or frame rate
|
| 59 |
+
|
| 60 |
+
## Nightingale Project
|
| 61 |
+
|
| 62 |
+
These models are part of the [Nightingale Video Enhancement Service](https://github.com/your-username/nightingale), an AI-powered video enhancement platform that provides:
|
| 63 |
+
|
| 64 |
+
- Web-based interface for video upload and processing
|
| 65 |
+
- Docker deployment for easy setup
|
| 66 |
+
- GPU acceleration with CPU fallback
|
| 67 |
+
- Real-time processing monitoring
|
| 68 |
+
- Batch processing capabilities
|
| 69 |
+
|
| 70 |
+
## License
|
| 71 |
+
|
| 72 |
+
The models retain their original licenses:
|
| 73 |
+
- Real-ESRGAN: BSD 3-Clause License
|
| 74 |
+
- RIFE: MIT License
|
| 75 |
+
|
| 76 |
+
This model repository packaging: MIT License
|