PocketGym

Progress Presentation 1
Team 9
Aakriti Bhandari
Kobe Hendrix
Evan Musick
Erdenesuren Shirmen
CSC 450 — Introduction to Software Engineering — Spring 2026

Software Backlog

  • 22 total backlog items across 3 sprints
    • Sprint 1 — Core Foundation: Pose estimation, exercise recognition, form analysis, rep tracking
    • Sprint 2 — User Management & Analytics: Authentication, profiles, dashboard, workout history
    • Sprint 3 — Extra Features: Exercise transitions, audio feedback, rest timers, progress charts
Overall Progress ~20%

Task Division

Each team member owns a primary domain across all sprints

Evan

  • Pose estimation pipeline
  • Camera & video processing
  • User preferences
  • Rest timers
  • Exercise library

Iggy

  • Exercise classification
  • ML model training
  • Workout history log
  • Audio form corrections

Kobe

  • Form analysis engine
  • Form scoring & tips
  • Post-workout summary
  • Exercise transitions

Aakriti

  • Rep & set tracking
  • User management
  • Account auth system
  • Dashboard & analytics

Completed Tasks

IDTaskStatus
PE-01 Camera capture & frame preprocessing Done
PE-02 MediaPipe pose estimation integration (33 keypoints) Done
PE-03 Skeleton overlay rendering with joint labels Done
PE-04 Joint angle calculation Done
— Full project scaffolding (models, APIs, views, templates) Done
— Sprint backlogs, documentation, and repo setup Done

Tasks to Be Completed

SprintTaskStatus
1 Exercise classification ML model In Progress
1 Form analysis & real-time scoring In Progress
1 Automatic rep counting & set tracking In Progress
2 User auth, profiles, dashboard & analytics Not Started
3 Audio corrections, rest timers, exercise library, charts Not Started

Challenges & Issues

Technical

  • MediaPipe pose accuracy varies with lighting and camera angle
  • Balancing real-time performance with detection accuracy
  • Training the exercise classification model requires curated datasets

Team / Process

  • Coordinating parallel development across ML, backend, and frontend
  • Integrating independently developed modules into a working pipeline
  • Keeping sprint pace with everyone's schedules

Live Demo

Pose Estimation & Skeleton Rendering

Next Steps

Finish Sprint 1 Complete angle calculation, exercise classification, form analysis, and rep counting
Integration Connect pose estimation output to classification, form analysis, and tracking modules
Sprint 2 Kickoff User authentication, profiles, workout history, and dashboard views
Testing Unit tests for each module and integration testing of the full pipeline
Questions?