← All projects

Smartphone Warranty Risk Assessment

  • Cracked-screen detection: Developed computer vision AI pipelines utilizing Meta's SAM3 to detect cracked smartphones from imagery for warranty triage, noting data scarcity and real-world data diversity.

    AI-detected crack regions on a smartphone back panel
    Example of cracked-screen detection

  • Drop detection and evaluation: Developed dynamics-aware algorithms to automatically detect and evaluate smartphone drops from onboard IMU data.

    3D visualization of phone orientation after a detected drop
    Example of post-drop orientation view

  • Battery health estimation: Developed a data- and energy-efficient on-device battery health estimation method aligned with constrained sensing, compute, and power budgets typical of handsets.

  • Drop orientation visualization: Developed a 3D visualization of the phone's final orientation after each detected drop to make incident outcomes easier to interpret for review and reporting.