Face-LLaVA transforms how AI interprets human facial expressions and attributes through a specialized multimodal large language model designed for face-centered applications.

• Enables in-context learning for facial expression and attribute recognition
• Generates natural language descriptions for facial reasoning
• Leverages the custom-built FaceInstruct-1M dataset for facial analysis
• Demonstrates strong performance in understanding facial communication

Medical Impact: Face-LLaVA opens significant possibilities for patient monitoring, pain assessment, and mental health diagnosis through advanced facial expression recognition, enabling more objective and consistent evaluation of emotional states.

Face-LLaVA: Facial Expression and Attribute Understanding through Instruction Tuning