Inspiration
Many of us may struggle with impatience and anger and that can have dire consequences, potentially causing us to blow up on friends and family and tarnish the relationships that we hold most dear. I built calmify to help keep us accountable and train us to stay calm and curb our anger. This app fosters healthy habits and encourages users to be calm and happy instead of angry and impatient. It uses machine learning to track the user’s emotions in order to give feedback about how the user is doing emotionally. The use cases are manifold: parents can use this to help train their patience with raising disobedient children; schools can use it to help teachers maintain their calm when students are being rowdy and frustrating; parents can use it for teenagers to discourage angry outbursts, and the list goes on. I think this webapp can benefit users by encouraging healthy habits which can help mend relationships and teach important values such as self control and patience and respect for one another.
What it does
This app is intended to be used regularly by users so that they can track their anger levels over time in an effort to improve on calmness and patience. It can be used for important purposes such as abuse detection and conflict de-escalation. From a technical standpoint, this app combines machine learning of video and audio input in order to give a holistic prediction of a user’s emotion, with anger being the emphasized emotion in question. The idea is that the user can track their aggregate (voice+visual) anger score over time and see how their anger management journey is progressing.
How we built it
Almost everything was significantly harder than expected. First of all, it took a very long time to decide on an idea. After that, it took many many hours to get the video analysis working. The audio analysis also took a considerable amount of time. The hardest part was needing to juggle everything by myself between the web development and machine learning aspects of the project. This was also my first time using python for a web app and making client side calls to the ML models which took a significant amount of time to figure out. There was also not a lot of online content available for emotion detection from audio that was suitable for my use case so it took a lot of research to find the appropriate resources.
Challenges we ran into
Building this by myself in a very short period of time.
Accomplishments that we're proud of'
A lot about machine learning and specifically using it in a web app.
What we learned
A lot about machine learning and specifically using it in a web app.
What's next for open bit
I want to improve on the machine learning techniques to get more accurate predictions. I also want to flesh out the scoring system as well as historical tracking of previous scores and eventually gamify the experience to give users extra incentive to improve on their anger issues.
Log in or sign up for Devpost to join the conversation.