Skip to content
pablo mendoza
← All projects

ShellHacks 2025 / 36-hour hackathon

LockedIn.

Computer vision for more focused work.

Real-time distraction tracking and useful session insights.

LockedIn / Project trailerExplore the build ↓

The build

A computer vision focus platform

Built in 36 hours at ShellHacks 2025 with Fernando, Tristan, and Gerrard. Our team connected webcam-based distraction detection with real-time feedback, session analytics, and social accountability.

My contribution

I contributed across the stack, helping integrate computer vision, backend services, and the dashboard into a working hackathon prototype.

  • Implemented WebSocket communication and integrated FastAPI endpoints for live updates.
  • Developed backend distraction analytics and activity metrics.
  • Worked on the React/Next.js dashboard and live focus visualizations.
  • Next.js
  • FastAPI
  • YOLOv8
  • MediaPipe
  • TensorFlow.js
  • WebSockets
LockedIn dashboard showing live phone-distraction detection, focus metrics, and a camera preview.
Live distraction tracking / Open image for a closer look ↗
The four-person LockedIn team posing in front of a ShellHacks backdrop surrounded by ocean-themed decorations and blue balloons.
The team behind LockedIn / ShellHacks 2025 ↗

From detection to feedback

From camera input to focus insightsSelect a step to explore

Camera-based focus tracking

Webcam

A webcam provides visual input during a study session, adding context beyond a countdown timer.

Input
Study-session camera input
Output
Visual activity for detection

Session analytics

Focus scores, distraction events, and timelines.

Social accountability

Friends and leaderboards for shared motivation.

Personal sessions

NextAuth accounts and Prisma-backed application data.

Conceptual information flow, not a deployment diagram. Vision tools are grouped without assigning individual models to execution environments.