RAPTOR AI Framework

<img src="RAPTOR_LOGO.png" alt="RAPTOR Logo" width="200"/> Rapid AI-Powered Text and Object Recognition Open-Source Content Insight Engine for Enterprise AI Applications <a href="https://github.com/DHT-AI-Studio/RAPTOR/blob/main/LICENSE"> <img src="https://img.shields.io/badge/License-Apache%202.0-blue.svg" alt="License"> </a> <a href="https://github.com/DHT-AI-Studio/RAPTOR/releases"> <img src="https://img.shields.io/badge/version-Aigle%200.1%20(Beta)-orange.svg" alt="Version"> </a> <img src="https://img.shields.io/badge/status-Beta-yellow.svg" alt="Status"> <img src="https://img.shields.io/badge/python-3.8%2B-blue.svg" alt="Python"> <a href="https://github.com/DHT-AI-Studio/RAPTOR/stargazers"> <img src="https://img.shields.io/github/stars/DHT-AI-Studio/RAPTOR?style=social" alt="GitHub stars"> </a> <a href="https://github.com/DHT-AI-Studio/RAPTOR/network/members"> <img src="https://img.shields.io/github/forks/DHT-AI-Studio/RAPTOR?style=social" alt="GitHub forks"> </a> <a href="#-about-raptor">About</a> • <a href="#-features">Features</a> • <a href="#-installation">Installation</a> • <a href="#quick-start">Quick Start</a> • <a href="#-documentation">Documentation</a> • <a href="#-contributing">Contributing</a> • <a href="https://dhtsolution.com/">Website</a>

RAPTOR is an advanced AI framework developed by the DHT Taiwan Team at DHT Solutions.

🚀 Current Release

Aigle 0.1 - First Community Beta Release

This is the initial open-source release of the RAPTOR framework, codenamed "Aigle". We're excited to share this with the community and look forward to your feedback and contributions.

🧪 Aigle 0.1 - Evaluation and Testing API

To help developers get started with the RAPTOR framework quickly and easily, we've deployed a test run API on DHT's development infrastructure. This evaluation API allows developers to:

Test and evaluate RAPTOR capabilities without setting up infrastructure
Develop AI applications using the RAPTOR framework with zero deployment overhead
Utilize DHT resources for testing and development purposes
Prototype faster by accessing pre-configured AI services

This is an excellent way to explore RAPTOR's features, build proof-of-concepts, and validate your use cases before deploying your own infrastructure.

🔗 Access the Evaluation API:
http://raptor_open_0_1_api.dhtsolution.com:8012/

For detailed API documentation, usage examples, and access instructions, please visit the link above.

Note: This is a development environment intended for evaluation and testing purposes. For production deployments, please refer to the Installation and Development sections below.

🎯 About RAPTOR

RAPTOR (Rapid AI-Powered Text and Object Recognition) is a Content Insight Engine that represents a paradigm shift in digital asset management, transforming passive media storage into an intelligent knowledge platform. By leveraging cutting-edge AI technologies including large language models, vector search, and semantic understanding, RAPTOR enables organizations to unlock the full value of their media assets through automated analysis, intelligent search, and actionable insights.

Business Value Proposition

85% reduction in manual content tagging and metadata generation
10x faster content discovery through semantic search
60% improvement in content reuse and operational efficiency
Real-time insights from video, audio, and document content
Enterprise-grade security, scalability, and integration capabilities

Strategic Differentiators

AI-Native Architecture: Built from the ground up around LLM orchestration and vector search
Multi-Modal Understanding: Unified analysis across video, audio, image, and text
Semantic Intelligence: Context-aware search that understands intent, not just keywords
Open + Enterprise Model: Open-source core with premium enterprise features
Production-Ready: Kubernetes-native with auto-scaling, fault tolerance, and 99.9% uptime

✨ Features

Version Aigle 0.1

This first community release includes:

Core Capabilities

Multi-Modal Content Analysis: Process and understand video, audio, images, and text
Semantic Search Engine: Context-aware search using vector embeddings
AI-Powered Metadata Generation: Automated tagging and content classification
LLM Orchestration: Flexible integration with multiple language models
Vector Database Integration: High-performance similarity search and retrieval

Intelligence Features

Content Understanding: Extract insights from unstructured media
Entity Recognition: Identify people, places, objects, and concepts
Sentiment Analysis: Understand emotional context in content
Topic Modeling: Automatic categorization and clustering
Temporal Analysis: Track content evolution over time

Enterprise Ready

Scalable Architecture: Kubernetes-native deployment
API-First Design: RESTful APIs for seamless integration
Security: Enterprise-grade authentication and authorization
Monitoring: Built-in observability and logging
Extensible: Plugin architecture for custom processors

For detailed release notes, see CHANGELOG.md.

📦 Installation

# Clone the repository
git clone https://github.com/DHT-AI-Studio/RAPTOR.git
cd RAPTOR/Aigle/0.1

# Create virtual environment (recommended)
conda create -n CIE python=3.10
conda activate CIE

# Install required dependencies:
pip install -r requirements.txt

Development

Step 1.

cd raptor
chmod +x check-services.sh deploy.sh logs.sh rollback.sh stop-all.sh tag-backup.sh
./deploy.sh

Step 2.Deploy service

Check container status
```
./check-services.sh
```

Test API connectivity

# Modellifecycle 服務
curl -s http://192.168.157.165:8086/docs

# Assetmanagement
curl -s http://192.168.157.165:8010/docs

View service logs
```
./logs.sh <service_name>
```

Quick Start

Create a new user and assign a new branch to the user

curl -X 'POST' \
'http://192.168.157.165:8086/users' \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-d '{
"username": "user1",
"password": "dht888888",
"password_hash": "",
"branch": "",
"permissions": [
    "upload",
    "download",
    "list"
]
}'

Create a new access token for the user

curl -X 'POST' \
'http://192.168.157.165:8086/token' \
-H 'accept: application/json' \
-H 'Content-Type: application/x-www-form-urlencoded' \
-d 'grant_type=password&username=user1&password=dht888888&scope=&client_id=string&client_secret=********'

Access RedisInsight 🔗 http://192.168.157.165:5540

Add a new connection:
- Connection Type: Redis Cluster
- Host: redis1
- Port: 7000
- Name: Redis Cluster
- Authentication: dht888888
Check if the local Ollama model includes qwen2.5:7b
```
ollama list
```
If not present
```
ollama pull qwen2.5:7b
```

curl -X 'POST' \
  'http://192.168.157.165:8010/models/register_ollama' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
  "local_model_name": "qwen2.5:7b",
  "model_params": 7,
  "registered_name": "qwenforsummary",
  "set_priority_to_one": false,
  "stage": "production",
  "task": "text-generation-ollama",
  "version_description": "Register qwen2.5:7b local model"
}'

Check if the registration was successful

curl -X 'GET' \
  'http://192.168.157.165:8010/models/registered_in_mlflow?show_all=false' \
  -H 'accept: application/json'

Start audio/video/image/document service
9.1 Create Kafka topics:

cd path/to/kafka
chmod +x create_topic.sh
sudo ./create_topic.sh

9.2 Starting Services

cd services
chmod +x start_services.sh
./start_services.sh

9.3 Check if all services are still running

./check_services.sh

Produce a test requests topic

cd /path/to/kafka/test_service
python test.py

View Service Logs
```
cd path/to/kafka
tail -f service_name.log
```
Available service names include: document_orchestrator_service, document_analysis_service, document_summary_service, document_save2qdrant_service, etc. You can replace document with audio, video, or image based on the required service, for example: audio_orchestrator_service, video_analysis_service, image_summary_service, etc

Check Redis Data

sudo docker exec -it redis-kafka_dev redis-cli --raw
GET "document_orchestrator:correlation_id"
GET "video_orchestrator:correlation_id"
GET "audio_orchestrator:correlation_id"
GET "image_orchestrator:correlation_id"

Note: Replace service_name and correlation_id with actual values.

Use the Qdrant Search API to query data
(8821/video_search, 8822/audi

RAPTOR

Install / Use

README