Inspiration

From a personal anecdote, germinated the inspiration for Yvette as it was noticeably a need that was neglected.

What it does

We decided to initiate and create a solution which translates in real time from sign language to text. In order to keep up with modern businesses we decided to have the solution be open source to the communities that require it.

How we built it

  • CNN: ML approach for classification
  • Keras/TensorFlow: Tools used for training the model.
  • RT Translation: We used opencv for the real time translation.
  • NLP: Neural Language Processing to give semantic sense to the structure.

Challenges we ran into

  • Idea: We have spent a moderate amount of time reviewing if the solution should be mainstreamed to casual users or if we should restrict it to B2B clients seeing as the former one is less practical considering the cost and the logistics required to make it work.
  • Technical aspect: We couldn't find APIs or models that turns around this idea, so we had to develop our own AI model prototype.
  • Business Plan: We had some trouble at first defining a profitable business plan. -Lack of data: we faced a crucial lack data of resources as much as we did with resources

Accomplishments that we're proud of

We are proud of everything we could accomplish in this short period of time. We learned many things along the way by facing the challenges we did. In fact, we could come up with a slick and modern identity chart compatible with the intentions and worldly nature of the solution. Group coordination was a little bit hard to manage at first considering the overlap or lack of in skills but we were able to overcome it after long sessions of brainstorming.

What we learned

How to learn and work in group and effectively plan out a solution and its implementation. Since we were very short on time, we had to brush up our time-management skills one way or another and all of that considering the lack of sleep.

What's next for Yvette

  • Including the different sign language dialects
  • Speech-to-text translation

Built With

Share this project:

Updates