-
-
Working of the Project
Inspiration
- In today’s hyper active world full of stress and pressure, mental and emotional health is becoming important factors in maintaining a person’s well-being, but with increasing work and social load, this gets neglected. Needless to say, it is important to address and take care of the mental and emotional health of individuals at homes, workplaces and schools.
- Nowadays, a number of social and medical institutions have come up to contribute towards helping victims of mental health disorders, spreading awareness and making active efforts in de-stigmatizing it.
- There are two major obstructions in the journey to heal mental health conditions. Firstly, lack of timely and proper diagnosis and treatment. Secondly, the stigma associated with mental health which makes it difficult for the victim to claim personal or professional help.
- However, some of the most preferred escape to mental health issues which people despite any age or gender seek is - spending time with oneself . So, I wanted to build something which can make a person feel much better when they are having their 'me-time' and help them transition into a better mood.
- Light and mood — two seemingly separate concepts that actually have a powerful and ongoing relationship with one another. There is an enormous amount of studies that showcase that ambient lighting and music have a major impact on the mood. This is also professionally use as color and sound therapy which is medically proven method to improve the mood of a person.
What it does
- The aim is to build a model capable of recognizing the person's emotions and transforming the ambiance of the room by changing the lights and playing music in accordance to the mood of the person, which can gradually help in pacifying their mood.
- This idea can be incorporated as a selfcare technique for personal use. With some modifications it can also be used in large scale like workplaces, restaurants and other public institution. For the hackathon, we implement the idea for personal use.
How we built it
- Emotion Detection - Deep Learning and Image Processing is used for detecting the emotion of a person. FER2013 dataset (available on Kaggle) was used to train a Convolutional Neural Network model. The model is capable of recognizing seven different mood of a person in real-time. However, we'll be using classification of three moods - Happy, Sad and Neutral.
- Ambiance Lighting - Once the model predicts the mood of a person, we try to set the ambiance of the room. Here, we have used Arduino UNO and WS2812B LED strip to demonstrate how the lights of the room changes according to the mood of the person.
- Playing Music - Simultaneously, the ambiance of the place can be improved by playing music which can improve the mood of the person. Here, we have a local database of songs we intend to play, which is fetched and played by VLC Media Player
Challenges we ran into
As a beginner in using Arduino, it took some time to figure out how to establish a communication channel between the Python program and the Arduino one. Also due to an imbalanced dataset, the validation accuracy was affected.
Accomplishments that we're proud of
Despite a low validation accuracy, the model is capable of classifying the emotions almost precisely in real-time. Working single handedly and completing the entire project within the deadline is something I'm very happy about!!
What we learned
I learned a lot about working with Arduino- something which was entirely new to me. As a beginner in Deep Learning, I got to experiment and learn more about optimizing the performance of the model.
What's next for Play Roshnee :)
- Looking forward to making the model more adaptive and the prototype more user friendly by adding some features of taking user preferences, including some other user info would help provide a personalized result.
- As of now, the prototype is ready and I wan to design and build it as a ready-to-be-used product.
Log in or sign up for Devpost to join the conversation.