SitRight

Demo

About

The problem it solves Studies show that over 70% of employees suffer from neck, shoulder, and back pain caused by poor sitting posture during long work hours. The problem often goes unnoticed until the discomfort becomes chronic—by then, damage to the spine, muscles, and joints may already have begun. Long periods of slouching, leaning forward, or craning the neck not only create physical strain but also lead to eye fatigue, reduced lung capacity, lower energy levels, and decreased productivity. SitRight AI is a smart, real-time posture monitoring app built to address this issue from the ground up. It continuously observes your sitting position through your device’s camera using advanced AI pose detection technology, running locally to ensure both speed and privacy. The moment it detects slouching, leaning, or any unhealthy sitting pattern, it responds instantly—alerting you through customizable voice alarms or discreet desktop notifications. By providing timely reminders to correct your posture, SitRight AI helps you maintain spinal alignment, reduce strain on muscles, and build healthier sitting habits over time. Whether you work in an office, study at home, or spend hours gaming, SitRight AI works silently in the background—protecting your posture so you can focus on what truly matters. ## ✨ Features - **AI Pose Detection** – Advanced pose estimation for accurate posture tracking. - **Customizable Alerts** – Choose between voice alarms or desktop notifications. - **Model Switching** – Optimized AI models for laptop or external webcams. - **Eye Strain Alerts** – Reminds you to rest your eyes at set intervals. - **Micro-Break Scheduler** – Encourages movement to avoid stiffness. - **Low-Latency Performance** – Runs locally for speed and privacy. - **Cross-Platform Ready** – Works as a Web App or Electron Desktop App. Smart Contract Address(es) 0xB060E44Ec364150c4E3B594A971C055D96e9a7De

How it's made

Challenges we ran into ** Challenges I Ran Into** - **Real-Time Performance without Lag** Achieving smooth, low-latency posture detection while keeping CPU usage minimal was a major hurdle. I had to fine-tune the detection interval (50ms) to strike the right balance between accuracy and performance, especially on devices with limited processing power. - Preventing Notification Spam Initially, the notification system would trigger repeatedly, causing a flood of alerts. I implemented a rate-limiting mechanism to ensure alerts only appear at fixed intervals, preventing “notification bombing.” - Multiple AI Model Integration Integrating two different Machine learning models (for laptop and external webcams) required restructuring the initialization logic so users could switch models seamlessly without restarting the app. - Cross-Platform Compatibility Ensuring the app worked as a packaged Electron desktop app for mac, linux and windows required careful handling of browser permissions, local file paths, and notification APIs. - UI and User Experience Balance Designing a simple yet functional UI that allowed quick switching between alert types, voice options, and models—without overwhelming the user—was an iterative process. - Maintaining Functionality When Unfocused Browsers often throttle background tabs, which stopped detection when the user switched windows. Moving to Electron solved this, but it required additional setup to keep loops running even when the app lost focus.