An open source Indian Number Plate Recogniton project built using deep learning.
-
24 August 21 - v1.0
- works with car number plates
python pip framework comming!!
-
Download or clone the repo
-
To run it locally, firstly install the requirements
pip instll requirements.txt
-
Then navigate to root folder from either file explorer or through cmd/terminal
-
In the root folder open cmd/terminal and type the following command
streamlit run app.py
- Then the app will be opened inside your browser.
Make sure you have chrome driver installed in your system and if you are using windows then go the file get_details.py change the path on line number 25,26 where your chorme drive is present.
chrome_path = r'path to chrome driver'
options.binary_location = r'path to chrome.exe'
If you are using linux then go to the get_details.py file and comment the lines 25,26,31 then uncomment lines 28 and 29.
options.binary_location = '/usr/bin/google-chrome'
driver = webdriver.Chrome(chrome_options=options,executable_path=r'/usr/bin/chromedriver')
To run it inside the docker container make sure that you have docker installed in your system. if you want to run the container locally then simply navigate to folder INPD_docker simply uncomment the following lines insider the dockerfile 77,81,83 and comment 78,82
EXPOSE 8501
#EXPOSE $PORT
ENTRYPOINT ["streamlit","run"]
#CMD streamlit run sapp_dock.py --server.port $PORT
CMD [ "sapp_dock.py"]
After making the changes you need to build the image using the folliwng command
docker build -t INPR-latest
Once the image has been built, then execute the following command
docker run -dp 3000:3000 INPR-latest
In few seconds go the url given below to access the app
http://localhost:3000
This is the initial build of the app so it won't be perfect in feteching the details of the number plate but to increase the success rate please make sure the uploaded image meets the following points.
1. The image should be clear and of high resolution
2. The number plate should also be visible clearly
3. The number plate should not be far away
4. The text on the plate should not be blured or distorted
5. The plate shold not be broken or partially covered
I am still trying to imporve it continiously so stay tuned!
Frontend: Streamlit
Backend: Python, Pytorch, Selenium
If you have any feedback, bugs, queries please reach out to me on Linkedin and also raise an issue about the same.