ScriptSense

KO | EN

gitlite — search

#python #hacktoberfest #react #go #golang #machine-learning #rust #java #windows #pytorch #security #mcp #ios #open-source

(★ 10)

Script analysis tool for extracting insights, understanding context, and improving written content.

ScriptSense 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

#  Language Identification & Script Analysis System
### Powered by AI4Bharat/Pralekha Dataset

An advanced NLP pipeline for **language identification**, **script recognition**, and **linguistic analysis** across 12 Indic languages + English using the [Pralekha dataset](https://huggingface.co/datasets/ai4bharat/Pralekha).

---

## 📦 Languages Supported

| Language | Script | Code |
|----------|--------|------|
| Bengali | Bengali | ben |
| English | Latin | eng |
| Gujarati | Gujarati | guj |
| Hindi | Devanagari | hin |
| Kannada | Kannada | kan |
| Malayalam | Malayalam | mal |
| Marathi | Devanagari | mar |
| Odia | Odia | ori |
| Punjabi | Gurmukhi | pan |
| Tamil | Tamil | tam |
| Telugu | Telugu | tel |
| Urdu | Perso-Arabic | urd |

---

## 🏗️ Architecture

```
lang_id_project/
├── src/
│   ├── data/           # Dataset loading, preprocessing, streaming
│   ├── models/         # ML models: char-ngram, transformer, ensemble
│   ├── analysis/       # Script analysis, Unicode inspection, statistics
│   ├── utils/          # Helpers, logging, metrics
│   └── api/            # FastAPI REST service
├── notebooks/          # EDA and experiment notebooks
├── configs/            # YAML configs
├── tests/              # Unit + integration tests
└── scripts/            # CLI tools
```

## 🚀 Quick Start

```bash
pip install -r requirements.txt

# Train all models
python scripts/train_pipeline.py --config configs/default.yaml

# Run API server
uvicorn src.api.server:app --reload --port 8000

# Analyze text
python scripts/analyze.py --text "नमस्ते दुनिया"

# Run full evaluation
python scripts/evaluate.py --split test
```

## 📊 Features

- **Multi-model ensemble**: Character n-gram + Transformer + Unicode heuristics
- **Script-level analysis**: Unicode block detection, script mixing, directionality
- **Streaming dataset pipeline**: Handles 1.5M+ rows with minimal RAM
- **FastAPI REST API**: Production-ready with async support
- **Detailed metrics**: Per-language F1, confusion matrix, confidence calibration
- **Interactive dashboard**: Gradio-powered web UI

ScriptSense

ScriptSense 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

ScriptSense

ScriptSense 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README