← Models

CV / YOLO26

YOLO26-seg

YOLO26 segmentation models on reComputer R Series with Hailo-8 or Hailo-10H.

136 downloads
Size
27.0 MB (Hailo-10H) / 33.7 MB (Hailo-8)
Memory
4GB+
Precision
Hailo HEF / HailoRT

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

Getting Started

Deploy
sudo docker run --rm \
  --name pi5-hailo8-yolo26m-seg \
  --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/yolo26m_seg:latest \
  python web_detection.py --model_path model/yolo26m_seg.hef --video_path video/test.mp4

Model Details

reComputer R Series (CM5 + Hailo-8)

YOLO26-seg on reComputer R Series (CM5 + Hailo-8)

YOLO26m-seg and YOLO26x-seg run on Hailo-8 through HailoRT 4.23. The one2one CPU decode (DFL + top-k + prototype-mask assembly) produces instance masks and boxes for 80 COCO classes.

The reComputer Hailo-8 repository currently provides compiled builds of yolo26m_seg and yolo26x_seg; n/s variants are not published there.

Model Info

VariantInputPackaged HEFContainer image
YOLO26m-seg640×640×3 RGB33.7 MByolo26m_seg
YOLO26x-seg640×640×3 RGB61.9 MByolo26x_seg

Hardware and Host Setup

ItemValue
BoardreComputer R Series with Raspberry Pi CM5
AcceleratorHailo-8 over PCIe, exposed as /dev/hailo0
RuntimeHailoRT 4.23.x (host / container must share major.minor)
Python in container3.11, aarch64
bash
sudo apt update
sudo apt install hailo-all -y
sudo reboot

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

# Docker
curl -fsSL https://get.docker.com -o get-docker.sh
sudo sh get-docker.sh --mirror Aliyun
sudo systemctl enable docker
sudo systemctl start docker

Run With Demo Video

Select the command matching the model variant.

YOLO26m-seg

bash
sudo docker run --rm \
  --name pi5-hailo8-yolo26m-seg \
  --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/yolo26m_seg:latest \
  python web_detection.py --model_path model/yolo26m_seg.hef --video_path video/test.mp4

Open http://<Board_IP>:8000 to view the web preview (instance masks overlaid per class + detection boxes).

USB Camera Mode

bash
sudo docker run --rm \
  --name pi5-hailo8-yolo26m-seg \
  --privileged \
  --net=host \
  -e PYTHONUNBUFFERED=1 \
  --device /dev/hailo0:/dev/hailo0 \
  --device /dev/video0:/dev/video0 \
  -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/yolo26m_seg:latest \
  python web_detection.py --model_path model/yolo26m_seg.hef --camera_id 0

YOLO26x-seg

bash
sudo docker run --rm \
  --name pi5-hailo8-yolo26x-seg \
  --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/yolo26x_seg:latest \
  python web_detection.py --model_path model/yolo26x_seg.hef --video_path video/test.mp4

REST API

bash
curl -X POST "http://<Board_IP>:8000/api/models/yolo26m_seg/predict" \
  -F "file=@test.jpg"
EndpointMethodPurpose
/api/models/yolo26m_seg/predictPOSTBoxes, classes, confidences (JSON)
/api/models/yolo26x_seg/predictPOSTBoxes, classes, confidences (JSON)
/api/video_feedGETMJPEG preview stream with masks
/api/configGET / POSTConfidence threshold
/api/video/upload + /api/video/analyzePOSTOffline video analysis

Development Notes

VariantSourceDockerfileContainer
YOLO26m-segsrc/rpi5_hailo8_yolo26m_seg/docker/hailo8/yolo26m_seg.dockerfileghcr.io/seeed-projects/recomputer-hailo8-cv/yolo26m_seg:latest
YOLO26x-segsrc/rpi5_hailo8_yolo26x_seg/docker/hailo8/yolo26x_seg.dockerfileghcr.io/seeed-projects/recomputer-hailo8-cv/yolo26x_seg:latest
  • The executor is the HailoRT 4.23 InferVStreams wrapper (one2one decode).
  • HEFs: Hailo Model Zoo v2.19.0 Hailo-8 builds.

Inputs and Outputs

Input: image (640x640 RGB). Output: instance masks, boxes, classes and confidences for 80 COCO classes.