Harpoon 1.1 SDK

Production-ready packages for seamless AI integration into defense systems

7 SDK Packages
3 Model Formats
5min Docker Deploy Time

SDK Packages

🐍

Python SDK

v1.1.0 • 15.2 MB

Complete Python integration package with ONNX and PyTorch model support. Includes optimized inference pipeline and GPU acceleration.

Includes:

  • ✅ harpoon-ai Python library
  • ✅ ONNX & PyTorch models
  • ✅ GPU acceleration support
  • ✅ Live streaming utilities
  • ✅ FastAPI server template
  • ✅ Docker deployment files

Requirements:

Python 3.8+, CUDA 11.8+ (optional)

C++ SDK

v1.1.0 • 32.1 MB

High Performance

Native C++ library for maximum performance. Ideal for embedded systems and real-time applications requiring minimal latency.

Includes:

  • ✅ Header-only library
  • ✅ ONNX Runtime integration
  • ✅ TensorRT optimization
  • ✅ Multi-threading support
  • ✅ CMake build scripts
  • ✅ Memory-efficient processing

Requirements:

C++17, CMake 3.16+, ONNX Runtime

🌐

REST API Server

v1.1.0 • 8.7 MB

Cloud Ready

Production-ready API server with authentication, monitoring, and WebSocket streaming. Deploy as microservice or container.

Includes:

  • ✅ FastAPI server application
  • ✅ JWT authentication
  • ✅ WebSocket streaming
  • ✅ Prometheus metrics
  • ✅ Docker & Kubernetes files
  • ✅ Load balancing support

Requirements:

Docker, Python 3.8+, 2GB RAM

📱

Edge SDK

v1.1.0 • 12.4 MB

Edge Optimized

Lightweight SDK optimized for ARM processors and edge devices. Quantized models for reduced memory footprint.

Includes:

  • ✅ ARM64 optimized binaries
  • ✅ Quantized INT8 models
  • ✅ Low-power mode support
  • ✅ Raspberry Pi examples
  • ✅ NVIDIA Jetson support
  • ✅ Offline deployment tools

Requirements:

ARM64, 1GB RAM, Linux kernel 4.14+

🐳

Docker Complete System

v1.1.0 • Ready to Run

Instant Deploy

Complete real-time drone detection system with webcam support and GUI. Includes green bounding boxes and multi-drone tracking. Zero setup required.

Features:

  • ✅ Real-time webcam detection
  • ✅ Green bounding box visualization
  • ✅ Multi-drone tracking (up to 4)
  • ✅ 12.2 FPS performance
  • ✅ 96.6% accuracy
  • ✅ Cross-platform Docker support

Requirements:

Docker, Webcam, X11 display (Linux/macOS/Windows)

🔗

Docker Model API

v1.1.0 • Lightweight

API Only

Lightweight REST API server for integrating drone detection into existing systems. No GUI, pure inference API with health monitoring.

API Endpoints:

  • ✅ POST /detect - Image detection
  • ✅ GET /health - Health check
  • ✅ GET /models - Model info
  • ✅ FastAPI with auto docs
  • ✅ JSON response format
  • ✅ Production ready

Requirements:

Docker, 500MB RAM, Port 8080

Standalone Models

Download individual model files for custom integration

ONNX Model

43.2 MB

Format: ONNX v1.12
Precision: FP32
Input: 640x640x3

PyTorch Model

21.8 MB

Format: PyTorch .pt
Precision: FP32
Framework: PyTorch 2.0+

TensorRT Engine

28.6 MB

Format: TensorRT 8.5
Precision: FP16
Optimized: RTX 30/40 series

Integration Tools

🔧

CLI Tool

Command-line interface for model testing, benchmarking, and deployment automation.

Testing Benchmarks Automation
📊

Monitoring Dashboard

Real-time performance monitoring with metrics visualization and alert management.

Metrics Alerts Real-time
🐳

Docker Images

Pre-built Docker containers for rapid deployment across any infrastructure.

Containers Kubernetes Scalable

Licensing & Usage

🆓 Development License

Free for evaluation, research, and non-commercial use. Includes all SDK packages with standard support.

  • Unlimited development use
  • Community support forums
  • Standard documentation

🏢 Commercial License

Production deployment with commercial support and service level agreements.

  • Production deployment rights
  • Priority technical support
  • Custom model training
  • SLA guarantees