gratheon/models-bee-detector

Microservice that detects bees

• Essentially Uses Ultralytics yolov5 model
• Runs as a http server
• Dockerized
• Uses weights by Matt Nudi https://github.com/mattnudi/bee-detection https://universe.roboflow.com/matt-nudi/honey-bee-detection-model-zgjnb

Architecture

Service diagram

flowchart LR
    web-app("<a href='https://github.com/Gratheon/web-app'>web-app</a>\n:8080") --> graphql-router("<a href='https://github.com/Gratheon/graphql-router'>graphql-router</a>") --> image-splitter("<a href='https://github.com/Gratheon/image-splitter'>image-splitter</a>\n:8800") --"send cropped image"-->models-bee-detector

Loading

Usage

# start service in cpu mode
just start

# alternatively, start in prod gpu mode
just start-jetson

CLI usage on bare host

This assumes you have installed all of the old dependencies

# webcam
python detect.py --weights yolov5s.pt --source 0

# video file
python detect.py --weights yolov5s.pt --source file.mp4

Ultralytics yolo v5 license

YOLOv5 is available under two different licenses:

• GPL-3.0 License: See LICENSE file for details.
• Enterprise License: Provides greater flexibility for commercial product development without the open-source requirements of GPL-3.0. Typical use cases are embedding Ultralytics software and AI models in commercial products and applications. Request an Enterprise License at Ultralytics Licensing.