https://github.com/akbartus/yolov8-segmentation-on-browser

Example of YOLOv8 Segmentation on Browser. It is powered by Onnx and served through JavaScript without any frameworks
https://github.com/akbartus/yolov8-segmentation-on-browser

computer-vision object-segmentation onnxruntime yolov8

Last synced: 2 days ago
JSON representation

Example of YOLOv8 Segmentation on Browser. It is powered by Onnx and served through JavaScript without any frameworks

• Host: GitHub
• URL: https://github.com/akbartus/yolov8-segmentation-on-browser
• Owner: akbartus
• License: mit
• Created: 2023-02-20T03:46:57.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-02-11T22:01:27.000Z (9 months ago)
• Last Synced: 2024-02-11T23:21:00.024Z (9 months ago)
• Topics: computer-vision, object-segmentation, onnxruntime, yolov8
• Language: JavaScript
• Homepage: https://yolov8-segmentation.glitch.me/
• Size: 16.2 MB
• Stars: 12
• Watchers: 1
• Forks: 4
• Open Issues: 1
• Metadata Files:
  • Readme: readme.md
  • License: LICENSE

Awesome Lists containing this project

README

        


  



This is adapted and rewritten version of YOLOv8 segmentation model (powered by onnx). This version can be run on JavaScript without any frameworks.

## Setup

To see it at work, just run index.html file. 

## Models

**Main Model**

YOLOv8n-seg model converted to onnx.

```

used model : yolov8n-seg.onnx

size       : ~ 13.5Mb

```

**NMS**

ONNX model to perform NMS operator [CUSTOM].

```

nms-yolov8.onnx

```

**Mask**

ONNX model to produce mask for every object detected [CUSTOM].

```

mask-yolov8-seg.onnx

```

## Use another model

It is possible to use bigger models converted to onnx, however this might impact the total loading time.

To use another YOLOv8 model, download it from Ultralytics and convert it to onnx file format. Read more on the [official documentation](https://docs.ultralytics.com/tasks/segmentation/#export)

**Custom YOLOv8 Segmentation Models**

Please update labels object inside of main.js file.

## Demo

To see demo, please visit the following page