We wanted to do an image recognition Chatbot that can name the location of a picture and suggest a link to Wikipedia of the place in question. However, this idea was too complicated to make it happen, so we finally decided to a text recognition Chatbot.
What it does
This chat bot translates words taken from a picture to text. From an input in English, it provides an output in any language available in Google Translate, as requested by the user. It can read the text in English written on the picture. For better results, the picture should be in black and white.
How we built it
The Chatbot was created on Cisco Spark platform and coded in python. The project was built using tesseract, Google translate API, NgRok and Github.
Challenges we ran into
We had to download a lot of modules to run the code. It was really hard to start working on the project since we didn't have much knowledge about how everything are connected to each other (AWS or NgRok).
Accomplishments that we're proud of
We are very proud to have a functional project.
What we learned
A simple idea does not mean it is easy to realize.
What's next for TranslatorBot
To make the use of this Chatbot more universal, we could improve the translation on colored background or the readability of stylized text. To make its use more accessible for visually impaired people, we could add an option to read the translated text out loud.