PicChat
Purpose and Inspiration:
Collecting correctly labelled data has always been a key to successful machine learning research. This project aims to encourage the public participation in providing well labelled data for future machine learning studies.This is achieved through a friendly UI, a widely accepted platform(facebook) and powerful ML API(IBM Watson). By demonstrating the effect of ML through friendly chatting with a bot, user is able to see the benefit of ML but also help the bot grow by providing correct label of custom uploaded images.
Chat with the bot at : https://www.facebook.com/CorgiChat/
Official Web page:http://cs.mcgill.ca/~shuang43/PicChat-ConduHacks/hack.html
Instructions:
-
Chat anything with Mr.Corgi and he will reply with the mood that you expressed
-
Upload Image in the chat and Mr.Corgi will recognize the objects in your image
-
Mr.Corgi can analyze your personality if you upload a biograph to him in the chat
Technologies Used:
-
HeroKu for hosting the backend
-
Facebook API for UI of the bot
-
IBM Watson visual recognition API
-
IBM Watson Tone analyzer
**NOTE: As the Facebook messenger bot is still under review, it may take at least 1 business day before it can be publicly used.