Design Guidelines for Continuous Vision AI (CVAI)

Taxonomy of pain points and design guidelines for CVAI pipelines

Design Guidelines for Continuous Vision AI (CVAI)

A comprehensive study on identifying and addressing pain points in Continuous Vision AI systems through systematic design guidelines and user studies.

Research Contributions

  • Taxonomy Development: Developed a comprehensive taxonomy of pain points across the entire CVAI pipeline, covering sensing, AI processing, and interaction components
  • Design Guidelines: Derived evidence-based design guidelines for building effective CVAI applications
  • Participatory Design: Organized design workshops to gather insights from domain experts and end users

Implementation

Built two functional Android prototypes to validate the design guidelines:

  1. Text Recognizer: Real-time text detection and recognition application
  2. Smart Yoga Teacher: Intelligent posture correction system using computer vision

Evaluation

  • Expert Review: Conducted thorough evaluation with domain experts in AI and mobile systems
  • User Study: Performed user studies to validate the effectiveness and usability of the design guidelines

Impact

The research provides practical guidance for researchers and practitioners developing continuous vision AI applications, addressing common challenges in sensing, processing, and user interaction.

Duration

June 2021 ~ December 2021