--- title: EdgeFirst AI emoji: ๐Ÿ”ฌ colorFrom: indigo colorTo: red sdk: static pinned: true license: apache-2.0 --- # EdgeFirst AI โ€” Spatial Perception at the Edge **EdgeFirst Perception** is an open-source suite of libraries and microservices for AI-driven spatial perception on edge devices. It supports cameras, LiDAR, radar, and time-of-flight sensors โ€” enabling real-time object detection, segmentation, sensor fusion, and 3D spatial understanding, optimized for resource-constrained embedded hardware. [![EdgeFirst Studio](https://img.shields.io/badge/EdgeFirst_Studio-3E3371?style=for-the-badge&logoColor=white)](https://edgefirst.studio) [![GitHub](https://img.shields.io/badge/GitHub-212529?style=for-the-badge&logo=github&logoColor=white)](https://github.com/EdgeFirstAI) [![Documentation](https://img.shields.io/badge/Documentation-1FA0A8?style=for-the-badge&logo=readthedocs&logoColor=white)](https://doc.edgefirst.ai) [![Au-Zone Technologies](https://img.shields.io/badge/Au--Zone_Technologies-6C757D?style=for-the-badge)](https://www.au-zone.com) --- ## Workflow EdgeFirst Model Zoo Ecosystem Every model in the EdgeFirst Model Zoo passes through a validated pipeline. [**EdgeFirst Studio**](https://edgefirst.studio) manages datasets, training, multi-format export (ONNX, TFLite INT8, eIQ Neutron, Kinara DVM, HailoRT HEF, TensorRT), and reference validation. Models are then deployed to our board farm for **full-dataset on-target validation** on real hardware โ€” measuring both accuracy (mAP) and detailed timing breakdown per device. Results are published here on HuggingFace with per-platform performance tables. ## Model Lifecycle Model Lifecycle: Training to Publication ## On-Target Validation On-Target Validation Pipeline Unlike desktop-only benchmarks, EdgeFirst validates every model on **real target hardware** with the full dataset. Each device produces both accuracy metrics (mAP) and a detailed timing breakdown โ€” load, preprocessing, NPU inference, and decode โ€” so you know exactly how a model performs on your specific platform. --- ## Supported Hardware ![NXP i.MX 8M Plus](https://img.shields.io/badge/NXP-i.MX_8M_Plus-3E3371?style=flat-square&logoColor=white) ![NXP i.MX 95](https://img.shields.io/badge/NXP-i.MX_95-3E3371?style=flat-square&logoColor=white) ![NXP Ara240](https://img.shields.io/badge/NXP-Ara240-3E3371?style=flat-square&logoColor=white) ![RPi5 + Hailo-8/8L](https://img.shields.io/badge/RPi5-Hailo--8%2F8L-1FA0A8?style=flat-square&logoColor=white) ![NVIDIA Jetson](https://img.shields.io/badge/NVIDIA-Jetson-76B900?style=flat-square&logoColor=white) --- ## Model Zoo Pre-trained YOLO models for edge deployment. Each model repo contains all sizes (nano through x-large), ONNX FP32 and TFLite INT8 formats, with platform-specific compiled variants as they become available. ### Detection | Model | Sizes | Nano mAP@0.5 | Link | |-------|-------|-------------|------| | **YOLO26** | n/s/m/l/x | 54.9% | [EdgeFirst/yolo26-det](https://huggingface.co/EdgeFirst/yolo26-det) | | **YOLO11** | n/s/m/l/x | 53.4% | [EdgeFirst/yolo11-det](https://huggingface.co/EdgeFirst/yolo11-det) | | **YOLOv8** | n/s/m/l/x | 50.2% | [EdgeFirst/yolov8-det](https://huggingface.co/EdgeFirst/yolov8-det) | | **YOLOv5** | n/s/m/l/x | 49.6% | [EdgeFirst/yolov5-det](https://huggingface.co/EdgeFirst/yolov5-det) | ### Instance Segmentation | Model | Sizes | Nano Mask mAP | Link | |-------|-------|--------------|------| | **YOLO26** | n/s/m/l/x | 37.0% | [EdgeFirst/yolo26-seg](https://huggingface.co/EdgeFirst/yolo26-seg) | | **YOLO11** | n/s/m/l/x | 35.5% | [EdgeFirst/yolo11-seg](https://huggingface.co/EdgeFirst/yolo11-seg) | | **YOLOv8** | n/s/m/l/x | 34.1% | [EdgeFirst/yolov8-seg](https://huggingface.co/EdgeFirst/yolov8-seg) | --- ## Naming Convention | Component | Pattern | Example | |-----------|---------|---------| | HF Repo | `EdgeFirst/{version}-{task}` | `EdgeFirst/yolov8-det` | | ONNX Model | `{version}{size}-{task}.onnx` | `yolov8n-det.onnx` | | TFLite Model | `{version}{size}-{task}-int8.tflite` | `yolov8n-det-int8.tflite` | | i.MX 95 TFLite | `{version}{size}-{task}.imx95.tflite` | `yolov8n-det.imx95.tflite` | | i.MX 93 TFLite | `{version}{size}-{task}.imx93.tflite` | `yolov8n-det.imx93.tflite` | | i.MX 943 TFLite | `{version}{size}-{task}.imx943.tflite` | `yolov8n-det.imx943.tflite` | | Hailo HEF | `{version}{size}-{task}.hailo{variant}.hef` | `yolov8n-det.hailo8l.hef` | | Studio Project | `{Dataset} {Task}` | `COCO Detection` | | Studio Experiment | `{Version} {Task}` | `YOLOv8 Detection` | ## Validation Pipeline | Stage | What | Where | |-------|------|-------| | **Reference** | ONNX FP32 and TFLite INT8 mAP on full COCO val2017 (5000 images) | EdgeFirst Studio (cloud) | | **On-Target** | Full dataset mAP + timing breakdown per device | Board farm (real hardware) | ## Perception Architecture | Layer | Description | |-------|-------------| | **Foundation** | Hardware abstraction, video I/O, accelerated inference delegates | | **Zenoh** | Modular perception pipeline over Zenoh pub/sub | | **GStreamer** | Spatial perception elements for GStreamer / NNStreamer | | **ROS 2** | Native ROS 2 nodes extending Zenoh microservices *(Roadmap)* | ## EdgeFirst Studio [**EdgeFirst Studio**](https://edgefirst.studio) is the MLOps platform that drives the entire model zoo pipeline. **Free tier available.** - Dataset management & AI-assisted annotation - Model training with automatic multi-format export and INT8 quantization - Reference and on-target validation with full metrics collection - CameraAdaptor integration for native sensor format training - Deploy trained models to edge devices via the [EdgeFirst Client](https://github.com/EdgeFirstAI/client) CLI --- Apache 2.0 ยท [Au-Zone Technologies Inc.](https://www.au-zone.com)