Metadata-Version: 2.4
Name: augvid
Version: 0.0.3
Summary: A collection of video augmentation layers
Author-email: Viktor Fairuschin <folio.peptisch-4q@icloud.com>
License-Expression: MIT
Project-URL: Repository, https://github.com/ViktorFairuschin/augvid
Keywords: augmentation,video,tensorflow,keras
Requires-Python: >=3.9
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: tensorflow>=2.12.0
Requires-Dist: numpy>=1.23.0
Provides-Extra: dev
Requires-Dist: pytest; extra == "dev"
Requires-Dist: opencv-python; extra == "dev"
Requires-Dist: matplotlib; extra == "dev"
Requires-Dist: decord; platform_system != "Darwin" and extra == "dev"
Requires-Dist: eva-decord; platform_system == "Darwin" and extra == "dev"
Dynamic: license-file

# AugVid

**AugVid** is a collection of augmentation layers for videos, inspired by the corresponding image preprocessing layers from `tf.keras`. 

<video src="https://github.com/user-attachments/assets/6ca99954-b8c5-4850-93ec-c30ecc640913"> demo </video>


## Installation

```bash
pip install augvid
```

## Getting Started

The augmentation layers can be added during the model construction:

```python
import tensorflow as tf
from augvid import RandomVideoBrightness, RandomHorizontalVideoFlip


model = tf.keras.Sequential([
    RandomVideoBrightness(max_delta=0.1),
    RandomHorizontalVideoFlip(),
    # add more layers here
])
```

## Demo

To generate demo video, first install the required dependencies:

```bash
pip install 'augvid[dev]'
```

Then run:

```bash
python demo.py --video <PATH_TO_VIDEO>
```
