ImageObjectCounter

upload image of articles and detect count in them based on trained model

At client (Reactjs)

1. run "npm i"
2. run "npm start"
3. open browser go to "localhost:1234"

At server (Python 2.7 & pip 18.0)

1. open cmd in main directory and run "pip install dlib-18.17.100-cp27-none-win32.whl --user"
2. run "pip install -r requirements.txt --user"
3. got server folder and run "server.py"

All Set To Go!!

To Train (- refer Trainer folder)

1. collect images of the item you want to train

2. put images in "/Trainer/data/" (name the folder with with item name)

like: "/Trainer/data/Parle_G" or "/Trainer/data/Amul_milk"

3. annotate the images with help of "labelImg.exe" to mark(draw rectangle around items) the images

*remember to select the PascalVOC as the save format

4. save the xml from labelImg in "/Trainer/data/"

same directory as its images

5. run "scrapper.py"
6. run "justTrain.py"
7. move the item folders to "server/data"