What it does
PulseAI is a web-based computer vision app which reads children's art for the emergence of emotional cues like anger, fear, sadness, or happiness. Through image classification, it shows strong emotions and lets the adult intervene appropriately when a child signals a need for support emotionally.
How we built it
We used a Convolutional Neural Network (CNN) which had been trained with a database of children's drawings which were labeled by emotions. The app includes:
Python
For the model, TensorFlow/Keras
Streamlit for the user interface
OpenCV for feature extraction from images
Challenges we faced
Collection and quality of the data set
Verifying correct emotion labeling
Making the app user-friendly for non-technical users
Image preprocessing and model compatibility management
Achievements
Trained an image classification model successfully
Deployed a working Streamlit app with an intuitive UI
Integrated smart interpretation of results for parents
What's next
Incorporate multilinguality
Provide a chat-based parent assistant (powered by LLMs)
Create PDF reports with interpretation and recommended next steps
Log in or sign up for Devpost to join the conversation.