CV / Tiny Yolo

Tiny-YOLOv3

reComputer R Series (CM5 + Hailo-10H)

COCO 80-class object detection with Tiny-YOLOv3 (CPU YOLOv3 decode) on reComputer R Series with Hailo-8 or Hailo-10H.

43 downloads

Size

8.4 MB

Memory

4GB+

Precision

Hailo HEF / HailoRT 4.23.x (Hailo-8) / 5.1.1 (Hailo-10H)

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 cm5-hailo10h-tiny-yolov3 \
  --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/tiny_yolov3:latest \
  python web_detection.py --model_path model/tiny_yolov3.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/tiny_yolov3/predict" \
  -F "file=@test.jpg"

Python

import requests

response = requests.post(
    "http://<Board_IP>:8000/api/models/tiny_yolov3/predict",
    files={"file": open("test.jpg", "rb")},
    timeout=30,
)
print(response.json())

Model Details

reComputer CM5 + Hailo-10H

Tiny-YOLOv3 on reComputer CM5 + Hailo-10H

Tiny-YOLOv3 performs COCO 80-class object detection on Hailo-10H through HailoRT 5.1.1. The HEF exposes 2 raw YOLO heads (13x13x255, 26x26x255); anchor decode and NMS run on the host CPU.

Model Info

Property	Value
Architecture	Tiny-YOLOv3 (COCO-80)
Task	Object detection
Input	416x416x3 BGR, aspect-preserving resize + pad 0 (mobilenet_ssd_ar)
Output	2 heads: 13x13x255 (stride 32), 26x26x255 (stride 16)
Parameters	8.85M
Operations	5.58G
mAP (COCO, full precision)	14.59% hardware reference
HEF	Hailo Model Zoo v5.4.0, Hailo-10H (7.1 MB)

Hardware and Host Setup

Item	Value
Board	reComputer R Series with Raspberry Pi CM5
Accelerator	Hailo-10H over PCIe, exposed as `/dev/hailo0`
Host driver	`hailo-h10-all` apt package
Runtime	HailoRT 5.1.1
Python in container	3.13, aarch64

bash

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

# After reboot
hailortcli fw-control identify
ls /dev/hailo0

Run With Demo Video

bash

sudo docker run --rm \
  --name cm5-hailo10h-tiny-yolov3 \
  --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/tiny_yolov3:latest \
  python web_detection.py --model_path model/tiny_yolov3.hef --video_path video/test.mp4

Open http://<Board_IP>:8000 to view the web preview.

REST API

bash

curl -X POST "http://<Board_IP>:8000/api/models/tiny_yolov3/predict"   -F "file=@test.jpg"

Implementation Notes

Anchors/strides verified against the Model Zoo v5.4.0 10H YAML: stride 32 {[81,82],[135,169],[344,319]}, stride 16 {[23,27],[37,58],[81,82]}.
Preprocessing: aspect-preserving resize + pad 0, raw BGR (the alls has no bgr_to_rgb and normalization is 0/255) — NOT the gray-114 RGB letterbox.
Executor is the HailoRT 5.1.1 multi-output FLOAT32 wrapper; first inference prints each head's value range per the SOP acceptance checklist.

Development Notes

Source module: src/hailo10h_tiny_yolov3/
Dockerfile: docker/hailo10h/tiny_yolov3.dockerfile
Container: ghcr.io/seeed-projects/recomputer-hailo10h-cv/tiny_yolov3:latest

Inputs and Outputs

Input: image, video, or USB camera frame. Output: COCO 80-class detection boxes with confidences, plus an annotated MJPEG preview.