Inspiration
The struggle to remember Names and as if remembering names wasn't difficult enough, the details of even the most recent conversation were fading from the memory. So something had to be done to avoid embarrassment and stress during these uncomfortable encounters.
What it does
Face Book is an application that is context aware, visually and auditory. It knows if you have a conversation with a friend and is able to check if he's already in your contacts. It remembers all the special details about your friends for you, so will never again forget a birthday or even struggle to find the perfect gift. Just because your contact book remembers that your friend is a Developer and likes Retro Games a lot.
How we built it
Planing the first draft of the general idea and a simple interface. By trial and error we were able to iterate from simple base functionality up to a first prototype that was able to understand if we were in a conversation with a friend or not. Fron there on we builded on top of it by iterating over remaining features from the idea pool of the first draft.
Challenges we ran into
Wishfull thinking, underestimating the complexity of the tasks are probably the biggest challenges we faced. After a really compelling pitch of the contact book we were overwhelmed by the creative ideas on what and how to implement, whilst maintaining a minimalistic designed Interface and have maximum convenience without losing usability and clarity. Prioritizing tasks was a challenge that was successfully overcome through good communication and teamwork.
The optimizing of the conversation state was until the end quite challenging. Just to separate two different voices by deterministic sound attributes has a big variety of possible combinations, thus finding an optimum is not trivial and highly iterative. Trusting the Process lead to stable results and the System is even able to distinguish between having a conversation with a Friend or a Stranger.
- Privacy issues: It is not legally allowed to just take a foto of somebody who might be unknown to you, so it becomes really important to find a measure that clarifies if the person is already in your contacts or at least a friend. Based on transscripted conversation snippets contact book is able to guess if you are talking to a known or unknown person. if the person is very likely to be known, contact book is able to compare via face recognition if this person is part of your contacts. If so its name pops up and a few important details and topics.
Accomplishments that we're proud of
Voice input based conversational awareness. Face Book is able to recognize different states based on audio input. The application stays in Idle state, when no one is speaking. From the first spoken words it is able to tell that someone is speaking or if there is a full grown dialogue it can even tell if you are friends from the content of your interaction.
Based on a friendliness ranking, the UI is able to distinguish different needs of the user. With a high friendliness score the UI will suggest to add a person to the contacts if that person is not already a friend.
What we learned
a lot...
What's next for [B23] Face Book
a more refined interface that is suggesting more robust different interactions based on the contextual awareness.
The face recognition to be able to check your contacts, so if you are talking to a friend, which already is in your contacts, there will be a fast face matching to bring up the most important informations immediately. For privacy reasons there will also be a consent popup to which the person in the frame has to actively consent with a voice command to enable sensitive data processing.
There will be a ranking of the most important topics as well as an update routine to keep your contact details always up to date. The overall goal is to process as many informations as possible on device, so that relation to your friends be in the spotlight and you'll never have to suffer from the agony that's being a mediocre friend.
Log in or sign up for Devpost to join the conversation.