We were sitting at dinner one day and were wondering if there could be an easier way to decide what to cook.
What it does
It analyzes a picture and identifies ingredients which are then used to look up recipes.
How we built it
We used a deep convolution neural net (DCNN) to extract the primary ingredient in each photo, store those ingredients, and fed them into a recipe search engine API and report back the top results.
Challenges we ran into
Originally we wanted to use object localization, but it was too challenging and we ended up using object classification. Also, a chatbot was intended to be the front end but because of difficult web communications and no server, we opted for a Flask based front end.
Accomplishments that we're proud of
The chatbot worked for about 3 questions.
What we learned
Downloading and learning a new piece of software in the set time is difficult.
What's next for JAKS Bot
Hopefully, integrating the chatbot and object detection.