← Back to projects
AI-Based Mental Health Screening System
PythonFlaskOpenCVNLTKTransformersLSTM
The Problem
Early detection of mental health indicators is challenging due to fragmented emotional and textual signals.
The Solution
Built an ensemble model combining Transformer architectures with LSTM networks to analyze facial emotions and textual sentiment simultaneously.
Architecture
Computer vision module (OpenCV) integrated with NLP pipeline, exposed via Flask-based API for real-time analysis.
Role
Machine Learning Developer
Impact
Improved prediction consistency by combining multimodal inputs, resulting in a more robust and scalable screening prototype.
Lessons I Learned
Multimodal AI systems significantly outperform single-input models in behavioral prediction tasks.