NE503 Verified Apps
The apps below have all been verified on real NE503 hardware. Download and install them directly for testing, or use the source code as a reference for your own development to accelerate POC delivery.
App List
| App | Scenario | Inference Model | Preview | Verification Result | Download |
|---|---|---|---|---|---|
| Person Detection | Perimeter intrusion / area intrusion detection | hailo_yolov8n_384_640 |    | Continuously detected 3 persons · 4.26M+ frames · ✅ Passed | app.yaml image.tar |
| Occupancy Monitor | Area people counting + occupancy monitoring | hailo_yolov8n_384_640 |   | Occupancy rate published periodically · frames 50→550 · ✅ Passed | app.yaml image.tar |
| Safety Helmet Detection | Construction safety PPE compliance — helmet / no-helmet detection with real-time alerts | safety_helmet_yolov8n_384_640 |    | Live detection of 9 persons (6 helmet + 3 no-helmet) · ✅ Passed | app.yaml image.tar |
Deployment
After downloading the app's app.yaml and image.tar, deploy via the Web Console:
Open Web Console → App Management → Import → Upload Package, upload app.yaml and image.tar separately, then click Install. After installation, click Start to launch the app.
Model preloading required
Each app depends on a different inference model. Before first use, load the corresponding model in Model Management on the Web Console:
- Person Detection / Occupancy Monitor: depend on the preinstalled
hailo_yolov8n_384_640— just scan and load it. - Safety Helmet Detection: depends on a custom-trained
safety_helmet_yolov8n_384_640(2-class Helmet / No Helmet; the app package does not include the model). Import the corresponding.hefinto Model Management and load it; the HEF can be obtained from the download link above or compiled from the app source.
More app use cases are being adapted. For custom app development, see the App Development Guide.
Document Version: v1.3 · Last Updated: 2026-07-21