The problem Your Moody Friend solves

With increasing screen time, the emotional and mental well being of people has been left on the sideline. Your Moody Friend will monitor the users' daily screen activities and provide an effective coping mechanism to make them sail through the difficulties with a companion who can understand them better.

Current Features:

1. A UI/UX that not only encaptivates the users but also makes their communication with the app comfortable as well as memorable
2. The mood tracker that keeps account of users' daily mood and observes the pattern it follows
3. The app will also help to keep a track of the amount of time user uses other apps and provide measures to keep the usage under check
4. Monitoring of parameters like voluntary and involuntary movements, heart beat fluctuations and temperature through mobile sensors
5. Using the analysis of users' Mood statistics, recommendation of activities like yoga, music, jokes, books, etc that can serve as effective coping mechanism)
6. An authentication module in the backend server, which will generate the pre-trained model for devices. Using tensorflow lite we tried to implement a small version of transfer learning where we planned to use data taken from device to further train the model on device

Future Extensions:

1. A Chat-bot that can serve as a perfect companion of the user
2. Reward system to celebrate and maintain performance streaks
3. Monitoring heart rate through face detection systems to further facilitate analysis
4. Integration of smart devices to measure physical parameters that can help in bridging the gap between body and mind, thereby paving the way towards improving all areas of well-being

Your Moody Friend is the go-to app whenever you're feeling low, wanting to talk to someone or searching for new ideas to spend your time productively.

Challenges we ran into

Major challenges we faced:

1. Creation of mood-dependent UI experience was daunting task considering possibilities of mental states a human mind can be in
2. Data privacy was a major concern when the app model was centralised. Therefore we planned to focus on Transfer Learning instead
3. Data collection for recommender system based on different set of users, classified on the basis of age, job, mood, etc didn't lead to satisfactory amount of data through surveys. So, we used various research papers for the purpose
4. Heart rate monitoring using face detection system was tedious and time-consuming process.