PROTOTYPE LINK: https://prime-vision-21.vercel.app/
PPT Link: https://1drv.ms/p/c/77f6ef183d598fba/EWDVVsYm4fxMuf2Hvao_tMsBbQZ9XpSwHogmom9IbwMAdg?e=vbr843
Demo Link: https://youtu.be/xXsiuB_ggx0
This is a web interface to YOLOv8 object detection neural network that allows to run object detection right in a web browser without any backend using ONNX runtime.
-
Click to any video or upload any video you want.
-
Click Play for object detection in video.
-
View results on Amazon E-Commerse website.
- HTML/CSS: Frontend interface for video upload, playback, and displaying detection results.
- JavaScript: Handles video processing, object detection, and communication with the Web Worker.
- YOLOv8: Implemented for real-time object detection in user-uploaded video content.
Clone this repository: git clone https://github.com/mohdhumaidd19/prime-vision-21
Download the last version of the "ort-wasm-simd.wasm" from here: https://cdn.jsdelivr.net/npm/onnxruntime-web/dist/ort-wasm-simd.wasm to the root of the repository.
You need to run index.html
using any local webserver, for example internal webserver of Visual Studio Code. Ensure that
the ONNX runtime library ort-wasm-simd.wasm
, the model file yolov8n.onnx
and the sample.mp4
video file exist in the same folder with index.html
.
Using the interface, you can press "Play" button to start object detection on the sample video.
- Upload a video: Use the "Upload Video" button to upload your video.
- Play the video: Watch the video to see real-time object detection in action.
- Interact with Objects: Click on the highlighted objects to view related products on Amazon