A web app which will detect the object for you and then translate it to any language of your choice :) Sounds Interesting? It is indeed interesting Pal!!.
1. Flask for building the backend of the web app.2. Tensorflow and it's API Keras for image processing and detection of image using the Resnet-152 Model which accurately detects the object with great accuracy.
3. Word2Word dataset for conversion of words to other languages.It is a dataset and an open-source Python package for cross-lingual word translations extracted from sentence-level parallel corpora. Developed by Korean Researchers mentioned in their 2019 Paper.
4. HTML, CSS, Js, Jquery for frontend of the web app.
- Let say you are in a foreign land and don't know what the object is in front of you. What will you do? You can use this application to detect that object and then translate it to the language of your choice :)
- Let say your kid doesn't know what that object is in the market. What he can do? Just use this web app, it will detect the object in no time with great accuracy .
- Make the whole website a working web app(I've made the service Worker file but it is not working, I guess.)
- Change the pre-trained model of Resnet into YOLOv3 which is the latest state-of-the-art for object detection by Google.
- Deploy the project on Heroku or DigitalOcean. I'm not planning to deploy it before I integrate at least one of the two features mentioned above. So hang tight for some time for the live demo :)
First of all, install all the python packages required and then install the weights from here and put this weights file in models directory(Create new one) in the root folder.
I'll be more than happy if you make a Pull request with the above features :)
<3 Open Source