Gesture-Based Navigation Techniques

I built this project to play Subway Surfers using nothing but hand gestures and computer vision. 🚄✋ Gesture-based interaction with computer vision is always a fun challenge. In this project, I combined OpenCV, MediaPipe, and PyAutoGUI to build a system where hand gestures are translated into real-time commands for applications and games. ⚙️ The implementation uses OpenCV and MediaPipe to detect and track face landmarks and hand movements. ⌨️ PyAutoGUI maps the recognized gestures into keyboard actions. 🌌 Selfie segmentation enables dynamic background replacement. 🕶️ Face mesh detection overlays virtual elements such as glasses, a hat, or a mustache in real time. 👉 The core idea is to transform simple hand signals into meaningful actions. The right hand can move right or up, while the left hand can move left or down. By combining gesture detection with visual overlays, the project creates an interactive experience that blends control and creativity. 🚀 This prototype shows how computer vision can support touchless interaction. The same methodology could be extended to education for interactive learning, to gaming for immersive controls, or to healthcare for hands-free operation. 🎥 The attached video demonstrates the system in action, highlighting the potential of gesture recognition in real-world applications. #ComputerVision #AI #MachineLearning #DeepLearning #GestureRecognition #OpenCV #MediaPipe #SubwaySurfers #Python

🎥 Exciting Project Showcase: Gesture-Controlled Presentation Demo 🎥 👋 Hey LinkedIn community! I'm thrilled to share a project I've been working on – a Gesture-Controlled Presentation Demo! 🤳📊 Ever wanted to navigate through a presentation with just a wave of your hand? With this program, I've harnessed the power of computer vision and hand tracking to create an intuitive and immersive way to interact with presentations. 🖐️✨ 🔍 How it works: 1. Swipe left ⬅️ (Extend your thumb) to seamlessly navigate back in your slides. 2. Swipe right➡️(Extend your pinky finger) to effortlessly move forward in your presentation. 3. Extend your index and middle fingers 🤞 to activate the pointer mode and emphasize key points. 4. Point and extend your index finger 👆 to draw on your slides, making annotations and explanations on the fly. 5. Spread all fingers 🖐️ to erase the drawn content and start anew. 💡 This project was a blend of programming, creativity, and practical application. By combining OpenCV, hand tracking, and thoughtful video editing, I've showcased a hands-on way to interact with presentations that break the mould of traditional slide navigation. 🌟 Whether you're a tech enthusiast, a presenter, or simply curious about innovative ways to interact with technology, I invite you to watch the video and try it yourself. Don't forget to share your thoughts and ideas in the comments below! Let's keep pushing the boundaries of what's possible in technology. 🚀💡 #GestureControlledPresentation #ComputerVision #HandTracking #Innovation #TechDemo #InteractiveTechnology #PresentationSkills

In my final research project for the MSc in Data Science program, I implemented Hand Gesture Recognition using both a custom CNN and MediaPipe. Through experimentation and evaluation, I demonstrated that MediaPipe significantly outperformed traditional methods in terms of speed, accuracy, and computational efficiency. Building on this foundation, I recently developed a Human Pose Estimation system where MediaPipe once again played a critical role by identifying and tracking 33 key body landmarks in real time. ○ Applications Across Projects: In the Hand Gesture Recognition project, MediaPipe was used to detect fingertip positions and joint angles. These were mapped to specific gestures (like thumbs up, open palm, etc.), which can be used to control a virtual interface or assist individuals with hearing/speech impairments. In the Human Pose Estimation project, MediaPipe helped identify posture and movement patterns. This is useful for fitness tracking, physical therapy, gaming, contactless interactions, and gesture-based learning tools. ○ Benefits of Using MediaPipe: ● No Need for Custom Dataset: MediaPipe uses pre-trained models, so there's no requirement for manual data collection or model training. ● Real-Time Performance: MediaPipe is optimized for real-time video processing, even on low-spec devices. ● High Accuracy: The models are trained on large-scale, high-quality datasets using deep learning, resulting in impressive accuracy even in fast movements or partial occlusions. ● Cross-Platform Compatibility: It supports Android, iOS, web, and desktop, which makes deployment more flexible for various applications like AR/VR, fitness apps, and educational tools. ● Easy Integration with OpenCV and Python: It works smoothly with OpenCV, making it easier to visualize results and build interactive applications. ○ Limitations & Suggestions for Improvement: ● Lighting Sensitivity: While MediaPipe works well under normal lighting, performance can drop in poor or inconsistent lighting. Improvement: Add adaptive brightness correction or preprocessing filters. ● Camera Dependency: Accuracy may vary based on webcam resolution or angle. Improvement: Use multi-angle support or implement camera calibration options. ● Single Person Tracking: By default, most MediaPipe models track one person at a time. Improvement: Explore models with multi-person support for collaborative or group-based applications. MediaPipe has greatly simplified access to high-level computer vision functionalities. Both of my projects became significantly more effective and accurate with it. Going forward, such frameworks are essential tools for building AI solutions that are smart, efficient, and societally impactful. #ComputerVision #MediaPipe #HumanPoseEstimation #GestureRecognition #AIInHealthcare