Vietnamese students face significant challenges in accessing mental health support due to cultural barriers and limited resources. To address this, we propose an AI-based solution that leverages advanced natural language processing techniques to enable personalized emotional support. In it, we build a model for emotion detection, demonstrating a promising F1-score of 0.6686 and a precision of 0.7718, highlighting its potential effectiveness in emotion recognition. The system provides an alternative to traditional chatbots, potentially enhancing mental health care for students in low-resource educational settings.