← Modèles

CV / License Plate

Tiny-YOLOv4 License Plates

Tiny-YOLOv4 License Plates detects plate regions in traffic images and video, using the official Hailo Model Zoo Hailo-8 HEF and CPU-side YOLO decoding.

4 téléchargements
Taille
8.21 MB
Mémoire
4GB+
Précision
Hailo HEF / HailoRT 4.23.x (Hailo-8) / 5.1.1 (Hailo-10H)

Choisissez l'appareil que vous utilisez. Le guide de configuration et la documentation seront mis à jour en conséquence.

Pour commencer

Déployer
sudo docker run --rm \
  --name cm5-hailo8-lp-detector \
  --privileged \
  --net=host \
  -e PYTHONUNBUFFERED=1 \
  --device /dev/hailo0:/dev/hailo0 \
  -v /usr/lib/libhailort.so.4.23.0:/usr/lib/libhailort.so.4.23.0:ro \
  -v /usr/lib/libhailort.so:/usr/lib/libhailort.so:ro \
  ghcr.io/seeed-projects/recomputer-hailo8-cv/tiny_yolov4_license_plates:latest \
  python web_detection.py --model_path model/tiny_yolov4_license_plates.hef --video_path video/test.mp4

API REST

Utilisez l'API REST pour exécuter l'inférence. Copiez les commandes ci-dessous.

Curl
curl -X POST "http://<Board_IP>:8000/api/models/tiny_yolov4_license_plates/predict" \
  -F "file=@traffic.jpg"
Python
import requests

response = requests.post(
    "http://<Board_IP>:8000/api/models/tiny_yolov4_license_plates/predict",
    files={"file": open("traffic.jpg", "rb")},
    timeout=30,
)
print(response.json())

Détails du modèle

reComputer R Series (CM5 + Hailo-8)

Tiny-YOLOv4 License Plates on CM5 + Hailo-8

Model information

This single-class detector uses the official Hailo Model Zoo v2.17 Hailo-8 HEF. It accepts a 416x416 RGB image and emits 13x13x18 and 26x26x18 raw heads. The application performs sigmoid activation, anchor/grid decode, thresholding, and NMS on the CPU.

The Model Zoo reports 74.083 mAP on Hailo's internal license-plate dataset. This metric is source data, not a local CM5 benchmark.

Demo video

The bundled 33.17-second, 1280x720 H.264 clip is transcoded from Hailo TAPPAS v3.29's official default LPR video:

https://hailo-tappas.s3.eu-west-2.amazonaws.com/v3.29/general/media/lpr_ayalon.mp4

It shows vehicles approaching a fixed elevated traffic camera with numeric plates visible throughout the sequence.

Run

bash
sudo docker run --rm \
  --name cm5-hailo8-lp-detector \
  --privileged \
  --net=host \
  -e PYTHONUNBUFFERED=1 \
  --device /dev/hailo0:/dev/hailo0 \
  -v /usr/lib/libhailort.so.4.23.0:/usr/lib/libhailort.so.4.23.0:ro \
  -v /usr/lib/libhailort.so:/usr/lib/libhailort.so:ro \
  ghcr.io/seeed-projects/recomputer-hailo8-cv/tiny_yolov4_license_plates:latest \
  python web_detection.py --model_path model/tiny_yolov4_license_plates.hef --video_path video/test.mp4

Open http://<Board_IP>:8000.

REST API

bash
curl -X POST "http://<Board_IP>:8000/api/models/tiny_yolov4_license_plates/predict" \
  -F "file=@traffic.jpg"

The response contains a license_plate confidence and pixel xyxy box for each detection.

Sources and validation

  • HEF: Hailo Model Zoo v2.17, Hailo-8 target.
  • Network facts: tiny_yolov4_license_plates.yaml from Model Zoo v2.17.
  • Video and official pipeline: Hailo TAPPAS v3.29.
  • Local status: syntax and content checks only; CM5 + Hailo-8 inference remains pending.

Entrées et sorties

Input: 416x416 RGB traffic frame. Output: license-plate confidence and xyxy box after raw-head decoding and CPU NMS.