Metadata-Version: 2.4
Name: tcasl
Version: 1.0.2
Summary: Python library for temporal contrast American Sign Language classification.
Author-email: Keshav Shankar <keshavshankar08@gmail.com>
Project-URL: Homepage, https://github.com/keshavshankar08/TCASL/tree/main/TCASLCore
Project-URL: Issues, https://github.com/keshavshankar08/TCASL/issues
Requires-Python: <3.11,>=3.10
Description-Content-Type: text/markdown
License-File: LICENSE.txt
Requires-Dist: opencv-python<4.11,>=4.9.0
Requires-Dist: numpy<2.0,>=1.26.0
Requires-Dist: torch>=2.0.0
Requires-Dist: torchvision
Requires-Dist: pillow>=9.0.0
Requires-Dist: lava-dl
Dynamic: license-file

# TCASL

TCASL is a lightweight, pure Python inference engine for predicting American Sign Language (ASL) gestures using Temporal Contrast, simulating a Dynamic Vision Sensor (DVS).

## Features

* **Zero-Bloat Inference:** A strictly defined PyTorch wrapper built specifically for rapid prediction.
* **Temporal Contrast Processing:** Built-in methods to convert standard webcam video into DVS-style event frames.
* **Auto-Formatting:** Automatically center-crops and down-scales raw video arrays to the 128x128 resolution required by the network.

## Installation

You can install the latest release of TCASL from PyPI using `pip`:

```bash
pip install tcasl
```

## Usage in Python

Examples of how the library can be used can be found in [examples/](https://github.com/keshavshankar08/TCASL/tree/main/TCASLCore/examples). You should not edit this code unless you read the documentation thoroughly, which is located at [src/tcasl/core.py](https://github.com/keshavshankar08/TCASL/blob/main/TCASLCore/src/tcasl/core.py).

## TCASL Project

You can find more information about TCASL on the [GitHub page](https://github.com/keshavshankar08/TCASL).
