ObjectDetectionExpress

This project takes trained model from Microsoft's Custom Vision and let's you self host it on your own server. Because it uses Tenserflow, it is really wobbly to run properly on a server. I recommend using included Dockerfile, bacause it is ready to run anywhere.

Requirements

Your CPU has to support AVX
You have to use Linux or WSL2 with kernel 5.10 or higher
You have to have installed python3 and C++ compilers

Installation

Install required packages

Debian based systems:

 sudo apt-get install build-essential wget python3 make gcc libc6-dev

Arch based systems:
```
 sudo pacman -S base-devel wget python
```
Other distros are not tested

Clone this repo and cd into it

git clone https://github.com/czM1K3/ObjectDetectionExpress
cd ObjectDetectionExpress

Install packages
```
yarn install
```
Run the server
- With hot reload
```
 yarn dev
```
- Without hot reload
```
 yarn start
```
Test it
1. Test in browser
  - Open browser and go to http://localhost:3000
  - Choose image and send it
2. Send POST message
  - Send multipart/form-data request to http://localhost:3000 with image
```
curl -X POST -F image=@/path/to/image http://localhost:3000/
```