COMP 491 Senior Design — Spring 2026
trAIn Logo

trAIn

An AI-powered mobile fitness coach that watches your form, counts your reps, and guides your fitness journey — all from your phone's camera.

Explore Features ↓ View on GitHub
5+
Exercises
30fps
Real-Time
ON-DEVICE
AI Inference

What Is trAIn?

trAIn is an Android fitness application designed for individuals who are beginners or have no experience with structured exercise. Using on-device computer vision and AI coaching, it guides users through bodyweight workouts with real-time form correction, automatic rep counting, and personalized post-workout feedback — making quality fitness coaching accessible to anyone with a smartphone.

The app combines pose estimation through MediaPipe, a motivational AI coach called Doctor Dopamine, and Firebase-backed progress tracking to create a complete training experience from onboarding to long-term improvement.

Profile creation via onboarding survey
Goal and motivation logging
Guided beginner-friendly workouts
Real-time form feedback via camera
AI-powered post-workout coaching
Workout history and progress stats
Workout guide with exercise tips
Private profile with long-term data

Beginners Deserve Better

Most new gym-goers face the same barriers: no guidance, no feedback, and no confidence. trAIn was built to change that.

🤷

No Starting Point

Beginners don't know which exercises to do, how to structure workouts, or where to begin their fitness journey.

👁️

No Form Feedback

Exercising alone means nobody corrects your form. Bad habits build silently, increasing injury risk.

📉

No Visible Progress

Without tracking or summaries, users can't see improvement — making it easy to lose motivation.

💸

Trainer Cost Barrier

Personal trainers are expensive. trAIn brings guided coaching to anyone with a smartphone.

What trAIn Does

A complete training pipeline: from onboarding to real-time coaching to post-workout analysis.

01

Smart Onboarding Survey

A short, beginner-friendly questionnaire captures age, weight, goals, and motivations to personalize every session.

User Experience
02

Guided Workout Selection

Curated bodyweight routines — push-ups, squats, lunges, crunches — tailored for beginners with clear instructions.

User Experience
03

Real-Time Pose Estimation

On-device computer vision tracks body landmarks at 30fps, enabling live rep counting and movement phase detection.

Computer Vision
04

Live Form Correction

Instant feedback on depth, alignment, posture, and range of motion. Corrective cues appear in real-time.

Computer Vision
05

Post-Workout Reports

Session summaries with rep totals, quality scores, fatigue estimates, and comparison data over time.

Data & Analytics
06

Doctor Dopamine AI Coach

A motivational AI persona delivers personalized recommendations, reflections, and encouragement tied to goals.

AI Coach

How The Vision System Works

From camera frame to coaching feedback in milliseconds — all running on-device for speed and privacy.

📷
Step 1
Camera Feed
CameraX streams live frames to the inference engine
🧠
Step 2
Pose Model
MediaPipe extracts 33 body landmarks per frame
📐
Step 3
Exercise Logic
Joint angles analyzed against exercise-specific rules
🔄
Step 4
Rep Detection
Phase changes trigger accurate rep counting
💬
Step 5
Live Feedback
Form corrections appear on-screen in real time
📊
Step 6
Summary
Aggregated metrics generate the post-workout report

Inside the App

Every screen designed to be readable at arm's length while exercising — minimal friction, maximum clarity.

👋

Welcome & Onboarding

Sets the tone with Doctor Dopamine and guides users into profile setup.

📋

Profile & Survey

Short form collecting age, weight, goals, and motivation preferences.

🏋️

Workout Selection

Browse beginner routines with clear descriptions and expected difficulty.

📸

Active Session

Camera view with live rep count, form feedback, and exercise progress.

📊

Post-Workout Report

Performance summary with quality scores, rep data, and AI coaching.

📈

History & Progress

View past sessions, track trends, and identify areas for improvement.

Replace emoji placeholders with actual app screenshots.

See trAIn In Action

A 3–5 minute walkthrough of trAIn's core features — from onboarding to a live workout session with AI coaching.

Upload your demo to YouTube, then replace this placeholder. <iframe src="https://www.youtube.com/embed/VIDEO_ID">

Built With

A modern Android-native stack combining Jetpack Compose, on-device ML, and cloud services.

🤖
Kotlin
Native Android development language
kotlinlang.org →
🎨
Jetpack Compose
Modern declarative UI framework
developer.android.com →
📷
CameraX
Android camera pipeline for live streaming
developer.android.com →
🧠
MediaPipe / MoveNet
On-device pose estimation & landmarks
ai.google.dev →
🔥
Firebase
Auth, Firestore, and user data storage
firebase.google.com →
🛠️
Android Studio
Primary IDE & emulator environment
developer.android.com →
✏️
Figma
UI/UX design and prototyping
figma.com →
🐙
GitHub
Version control & team collaboration
github.com/tabee1024/trAIn →

Skills Used

The technical and design competencies our team brought together to build trAIn.

Kotlin ProgrammingAndroid DevelopmentJetpack Compose UI/UX DesignComputer VisionPose Estimation Machine LearningFirebase / FirestoreUser Authentication REST API IntegrationMVVM ArchitectureReal-Time Data Processing Git & Version ControlAgile CollaborationFigma Prototyping Prompt EngineeringCamera APIsDatabase Schema Design

Doctor Dopamine

🧬
AI Training Persona

Your Personal Coach

Doctor Dopamine is the AI personality behind every motivational message, post-workout reflection, and goal-aware recommendation in trAIn. More than a chatbot — it's a training companion that remembers your goals and celebrates your progress.

"Great work today — your squat depth improved 15% since last Tuesday. Your consistency this week is building real strength. Tomorrow, try adding one extra set of lunges. You've got this."

Who Built trAIn

Six CSUN Computer Science seniors combining frontend, computer vision, backend, database, and AI expertise.

Christian Estrada

Christian Estrada

Team Lead · Implementation Support
Jesus Ramirez

Jesus Ramirez

Backend · AI Coach · Pose Logic
Tabitha Sulaiman

Tabitha Sulaiman

Front-End Development
Vishal Chaudhari

Vishal Chaudhari

Front-End · Back-End
Brian Bonilla

Brian Bonilla

Back-End · Live Feed Camera
Reenu Mohan

Reenu Mohan

Front-End Development · UI/UX
// source code
github.com/tabee1024/trAIn →

COMP 491 · Senior Design Project II — Spring 2026

California State University, Northridge

Department of Computer Science