CV / Face Landmarks

Face Landmarks Lite

Face Landmarks Lite detects 98 facial keypoints with contour connections through HailoRT inference, ideal for face tracking and AR applications.

2 downloads

Size

1.8 MB

Memory

4GB+

Precision

Hailo HEF / HailoRT

Choose the device you're using, the set up guide and documentation will update accordingly.

Platform

Inference Engine

Getting Started

Deploy

sudo docker run --rm \
  --name pi5-hailo-face-landmarks \
  --privileged \
  --net=host \
  -e PYTHONUNBUFFERED=1 \
  --device /dev/hailo0:/dev/hailo0 \
  -v /usr/lib/libhailort.so.5.1.1:/usr/lib/libhailort.so.5.1.1:ro \
  -v /usr/lib/libhailort.so:/usr/lib/libhailort.so:ro \
  ghcr.io/seeed-projects/recomputer-hailo10h-cv/face_landmarks_lite:latest \
  python web_detection.py --model_path model/face_landmarks_lite.hef --video_path video/test.mp4

REST API

Use the REST API to run inference. Copy the commands below.

Curl

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

Python

import requests

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

Model Details

Face Landmarks Lite on reComputer CM5 + Hailo-10H

MediaPipe Face Landmarks Lite (0.6M params, 972 FPS) for facial keypoint detection, running on Hailo-10H via HailoRT.

Model Info

Property	Value
Architecture	MediaPipe Face Landmarks
Task	Facial Landmark Detection
Input	192×192×3 RGB
Output	98 facial keypoints
Parameters	0.6M
FPS	972
Source	Hailo Model Zoo

Hardware and Host Setup

Item	Value
Board	Raspberry CM5
Accelerator	Hailo-10H, `/dev/hailo0`
OS	Raspberry Pi OS Bookworm, aarch64

bash

sudo apt update && sudo apt install hailo-h10-all -y
sudo reboot

hailortcli fw-control identify
ls /dev/hailo0

Run With Demo Video

bash

sudo docker run --rm --privileged --net=host \
  -e PYTHONUNBUFFERED=1 \
  --device /dev/hailo0:/dev/hailo0 \
  -v /usr/lib/libhailort.so.5.1.1:/usr/lib/libhailort.so.5.1.1:ro \
  -v /usr/lib/libhailort.so:/usr/lib/libhailort.so:ro \
  ghcr.io/seeed-projects/recomputer-hailo10h-cv/face_landmarks_lite:latest \
  python web_detection.py --model_path model/face_landmarks_lite.hef --video_path video/test.mp4

USB Camera Mode

bash

sudo docker run --rm --privileged --net=host \
  -e PYTHONUNBUFFERED=1 \
  --device /dev/hailo0:/dev/hailo0 \
  --device /dev/video0:/dev/video0 \
  -v /usr/lib/libhailort.so.5.1.1:/usr/lib/libhailort.so.5.1.1:ro \
  -v /usr/lib/libhailort.so:/usr/lib/libhailort.so:ro \
  ghcr.io/seeed-projects/recomputer-hailo10h-cv/face_landmarks_lite:latest \
  python web_detection.py --model_path model/face_landmarks_lite.hef --camera_id 0

HEF Files

HEF	Size	Notes
`face_landmarks_lite.hef`	1.8 MB	Facial landmark detection

REST API

text

POST http://<Board_IP>:8000/api/models/face_landmarks_lite/predict

bash

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

Endpoint	Method	Purpose
`/`	GET	Web preview UI
`/api/models/face_landmarks_lite/predict`	POST	98 keypoints (JSON)
`/api/video_feed`	GET	MJPEG preview stream

Development

Source: src/hailo10h_face_landmarks_lite/
Dockerfile: docker/hailo10h/face_landmarks_lite.dockerfile

Inputs and Outputs

Input: image (192x192 RGB). Output: 98 facial keypoints with confidence.