How to integrate an edge impulse model on your web app?

Help
#1

hello community!

i want to integrate my edge impulse model into an application, say one that’s built with Streamlit, how do i go about it?
i watched the video on “getting started with edge impulse SDK” but it doesn’t quite cover my case because the model was built on edge impulse and i am just looking to do inference with the model and display some output on an application interface

i have edgeimpulse package installed and the API jey for the project stored, please let me know how i can go about it or point me to useful resources

thanks!

#2

Sounds like you need to deploy as WebAssembly:

image
image600×735 66.4 KB

#3

hello, thanks for the help!
i downloaded the web assembly build and ran server.py in the browser folder and:

  • it seems like it is for a classification use case, i plan to deploy an object detection model
  • most of the code is written in js and i don’t know js

can’t i just save my model and provide the path to it or use the API key somehow in the backend of a live inferencing interface?

#4

server.py handles FOMO miodels. For example an image with 3 screws in it returned:

{
    "anomaly": 0,
    "results": [
        {
            "label": "screw",
            "value": 0.99609375,
            "x": 56,
            "y": 32,
            "width": 8,
            "height": 8
        },
        {
            "label": "screw",
            "value": 0.99609375,
            "x": 40,
            "y": 48,
            "width": 8,
            "height": 8
        },
        {
            "label": "screw",
            "value": 0.99609375,
            "x": 56,
            "y": 64,
            "width": 8,
            "height": 8
        }
    ]
}
  • If you know how to program you should be able to interpret what ChatGPT tells you as you learn a new language.