Is it possible to deploy my project on my server?

Help
#1

Hello !
I am working on a project which is to take photo of insects and recognize them, it requruired to ran on the phone. I’ve trired some solutions but the edge-impulse is better.
So I am wongdering if it can be deployed on a server running spring-boot, then I can use the App to take a photo and send back to the server, and my server can parse the photo to generate its features. Finally, the server recognize it and return the results to the App in json format.

I hava read the tutorials about running impulse on Linux ,node.js and web. But I found that the are all required to generate features through the edge-impulse website, then input the features into terminal.

Now I want to know its possibility, and how to generate the features of the photo uploaded without using the API of edge-impulse, also which is the best way to fullfill my requirement in the following three ways( On your desktop computer\ Through WebAssembly (Node.js)\ Through WebAssembly (browser)

or if there is a solution which can help me

I’ll be really appreciate your help!
And I’m sorry if my English is hard to understand!

#2

Hi @zhongrui yeah you can do this in two ways:

  1. Use the custom data examples in Node.js, Go, C++ or Python in Edge Impulse for Linux and use it to build a webserver that can classify features.
  2. Use Node.js + WebAssembly to do the same.

The features need to be in a very simple format: just one pixel per value of your image (in the resolution the model expects) - see https://docs.edgeimpulse.com/docs/running-your-impulse-locally-1#signal-layout-for-image-data for the image format. If you know the language of your server we can help you refine this even more!

1 Like
#3

Hi!
My server is running on Linux system, with Java language, and Spring-Boot structure. But I see most of the tutorials are compiled with C++, it may confuse me…

#4

Sorry for the late reply, but easiest might be to compile in C++, or use the Node.js or Pythono example, and then just invoke the script from Java (via exec() or something).

#5

got it, thanks for your help!

1 Like