Build and deploy high-performance AI applications on edge devices with Hailo hardware accelerators. From real-time computer vision to GenAI voice-to-action agents - production-ready applications and reusable infrastructure to accelerate your AI development.
Highlight: Voice-to-action AI agents that understand natural language commands and control hardware in real-time - elevators, servos, LEDs, and more.
Supports: Hailo-8, Hailo-8L, and Hailo-10H accelerators
Supported versions:
- Hailo-8 / Hailo-8L: HailoRT 4.23, TAPPAS Core 5.1.0
- Hailo-10H: HailoRT 5.1.1 & 5.2.0, TAPPAS Core 5.1.0 & 5.2.0, GenAI Model Zoo 5.1.1 & 5.2.0
💡 Running Multiple Apps: You can run multiple Hailo applications in parallel (e.g., Vision + GenAI). See the Running Parallel Applications Guide for configuration details.
Perfect for: Raspberry Pi 5, x86_64 Ubuntu systems, and edge AI deployments.
✨ 20+ Ready-to-Run Applications
- 🎯 GenAI - Voice to Action (Featured): AI agents that convert natural language into hardware control - talk to your devices and make things happen
- Computer Vision: Object detection, pose estimation, instance segmentation, face recognition, depth estimation, OCR
- More GenAI: Voice assistants, VLM chat, speech recognition (Whisper)
- Advanced: Multi-camera tracking, zero-shot classification (CLIP), tiling for high-res processing
🚀 Production-Ready Infrastructure
- GStreamer-based pipelines for efficient video processing
- Modular Python library for rapid prototyping
- Hardware-optimized performance with Hailo accelerators
🛠️ Developer-Friendly
- Install in minutes with automated scripts
- Extensive documentation and examples
- Easy integration into your own projects
Learn more: Hailo Official Website | Community Forum
Before installing hailo-apps, ensure you have the following Hailo packages installed:
- hailort PCIe driver (.deb package)
- hailort (.deb package)
- hailo-tappas-core (.deb package)
- hailort Python wheel
- hailo_tappas_core_python_binding Python wheel
Note: These packages are available from the Hailo Developer Zone. See the Full Installation Guide for detailed setup instructions.
git clone https://github.com/hailo-ai/hailo-apps.git
cd hailo-apps
sudo ./install.shNeed help? See the Full Installation Guide for hardware setup and troubleshooting.
source setup_env.sh # Activate environment
hailo-detect-simple # Start object detection
Try more computer vision:
hailo-pose # Pose estimation
hailo-seg # Instance segmentation
hailo-tiling # Tiling for high-res processing
hailo-depth # Depth estimation
Hailo-10H generative AI demos: voice assistants, VLM chat, voice-to-action agents
Supported versions:
- Hailo-8 / Hailo-8L: HailoRT 4.23, TAPPAS Core 5.1.0
- Hailo-10H: HailoRT 5.1.1 & 5.2.0, TAPPAS Core 5.1.0 & 5.2.0, GenAI Model Zoo 5.1.1 & 5.2.0
# Voice Assistant
cd hailo_apps/python/gen_ai_apps/voice_assistant/
python voice_assistant.py
# Agent Tools - Voice to Action
python -m hailo_apps.python.gen_ai_apps.agent_tools_example.agent
# VLM Chat - Vision + Language
cd hailo_apps/python/gen_ai_apps/vlm_chat/
python vlm_chat.pyAll GenAI Apps — hailo_apps/python/gen_ai_apps
| APP | Description |
|---|---|
voice_assistant |
Voice assistant with speech recognition and TTS |
agent_tools_example |
Voice-to-action agent tools: natural language → hardware control |
vlm_chat |
Vision-Language chat: combine images and text for reasoning |
simple_llm_chat |
Minimal text-only LLM chat example |
simple_vlm_chat |
Minimal VLM chat example (image + text) |
simple_whisper_chat |
Minimal Whisper-based speech recognition chat |
hailo_ollama |
Ollama integration utilities for running local LLMs |
Real-time pipelines for cameras, RTSP streams, and multi-source processing
All Pipeline Apps — hailo_apps/python/pipeline_apps
| APP | Description |
|---|---|
detection |
Real-time object detection pipeline |
detection_simple |
Lightweight object detection example |
instance_segmentation |
Instance segmentation pipeline |
pose_estimation |
Human pose estimation pipeline |
depth |
Depth estimation pipeline |
face_recognition |
Face detection and recognition |
tiling |
High-resolution tiling-based inference |
multisource |
Multiple camera/source pipeline |
reid_multisource |
Multi-source person re-identification |
paddle_ocr |
OCR pipeline (PaddleOCR) |
clip |
Zero-shot classification with CLIP |
Learn HailoRT with hands-on Python and C++ demos
All C++ Standalone Apps — hailo_apps/cpp
| APP | Description |
|---|---|
classification |
Image classification with models trained on ImageNet |
depth_estimation |
Depth estimation using scdepthv3 and stereonet |
instance_segmentation |
Instance segmentation with yolov5_seg and yolov8_seg |
object_detection |
Generic and asynchronous object detection |
onnxruntime |
Inference with Hailo device and postprocessing via ONNXRuntime |
pose_estimation |
Pose estimation with yolov8 |
semantic_segmentation |
Semantic segmentation with Resnet18_fcn (Cityscapes dataset) |
zero_shot_classification |
Zero-shot classification with clip_vit_b_32 |
All Python Standalone Apps — hailo_apps/python/standalone_apps
| APP | Description |
|---|---|
object_detection |
Object detection and tracking with YOLO, SSD, and CenterNet |
instance_segmentation |
Instance segmentation with yolov5_seg and yolov8_seg |
lane_detection |
Lane detection using UFLDv2 |
pose_estimation |
Pose estimation with yolov8 |
speech_recognition |
Automatic speech recognition with the Whisper model |
super_resolution |
Super-resolution with espcnx4 and SRGAN |
| Guide | What's Inside |
|---|---|
| User Guide | Installation, running apps, configuration, repository structure |
| Developer Guide | Build custom apps, write post-processing, model retraining |
This repository provides three types of applications, each suited for different development scenarios:
| App Type | Best For | Examples |
|---|---|---|
| Pipeline Apps | Production-ready video processing with cameras, RTSP streams, and real-time inference | Object detection, pose estimation, instance segmentation, face recognition |
| Standalone Apps | Learn HailoRT (Python/C++); install only specific apps; images/video/camera | Object Detection, OCR, Stereo Depth Estimation |
| GenAI Apps | Hailo-10H generative AI applications | Voice assistants, VLM chat, voice-to-action agents, Whisper speech recognition |
Use pipeline apps when you need real-time video processing with:
- Raspberry Pi Camera, USB cameras, or video files
- RTSP streams for IP cameras
- Multi-camera and multi-source processing
- GStreamer-based efficient pipelines
Use standalone apps when you need to:
- Install only specific apps; no TAPPAS required
- Learn the HailoRT API in Python/C++ (hands-on demos)
- Work with images, videos, and camera streams
- Prototype quickly and validate models
Use GenAI apps for Hailo-10H generative AI features:
- Voice assistants with speech recognition
- Vision-Language Models (VLM) chat
- Voice-to-action AI agents
- Whisper-based speech recognition
📚 See all applications: Running Applications Guide
🎯 Input Flexibility
- Raspberry Pi Camera, USB cameras, video files, RTSP streams
- Multi-camera and multi-source processing
⚡ Optimized Performance
- Hardware-accelerated inference on Hailo devices
- Efficient GStreamer pipelines for real-time processing
- Minimal CPU overhead
🧩 Modular & Extensible
- Reusable Python library (
hailo_apps) - Custom model support with HEF files
- Easy integration into existing projects
→ Learn about the repository structure
💬 Hailo Community Forum - Get help, share projects, connect with other developers
🐛 Issues? Search the forum or open a discussion - the community and Hailo team are here to help!
License: MIT - see LICENSE for details
