This repository contains a collection of real-time computer vision applications built using OpenCV, MediaPipe, and modern human–computer interaction techniques. The projects focus on gesture recognition, object tracking, interactive vision systems, and real-time augmented interfaces.
These applications are designed to run with a webcam and emphasize low-latency real-time processing suitable for research prototypes and interactive systems.
This portfolio demonstrates:
- Real-time color segmentation and centroid tracking
- Gesture-based human–computer interaction
- Interactive vision games and drawing systems
- Motion detection and surveillance pipelines
- Augmented reality visual effects
- Virtual musical instruments
- Gesture-controlled dataset navigation
- Virtual mouse and game control
- YOLO real-time object detection
- C++
- Python
- OpenCV
- MediaPipe
- NumPy
- PyAutoGUI
- YOLO (You Only Look Once)
- HSV color segmentation
- Image moments and centroid computation
- Background subtraction
- Gesture recognition
- Collision detection
- Coordinate mapping
- Real-time UI composition
- Motion tracking
- Augmented reality overlays
Real-time detection of red regions using HSV color thresholding. Dual HSV ranges handle hue wrap-around. Binary masking and morphological filtering isolate red pixels.
Key techniques
- HSV color segmentation
- Thresholding
- Morphological noise removal
Applications Color-based object tracking and gesture detection.
Extends red detection by computing the barycenter of detected pixels using image moments. The centroid is tracked and visualized in real time.
Key techniques
- Image moments
- Centroid computation
- Real-time motion tracking
A gesture-controlled interactive game where fingertip tracking is used to pop balloons. Includes scoring, animations, sound effects, and a leaderboard.
Key techniques
- Hand landmark detection
- Collision detection
- Game state management
- Real-time gesture interaction
Tracks centroid motion to simulate a virtual paintbrush. Color palettes and a clear button enable gesture-based drawing on a live video canvas.
Key techniques
- Centroid motion tracking
- Interactive UI overlays
- Path rendering
Adaptive background modeling detects large motion events. Intrusions trigger alarms and automatically save timestamped snapshots.
Key techniques
- Background subtraction
- Contour filtering
- Real-time alert systems
Creates an augmented reality invisibility illusion by replacing cloak-colored pixels with a stored background.
Key techniques
- Color masking
- Background replacement
- Image compositing
A gesture-controlled musical interface where fingertips trigger notes mapped to on-screen piano keys.
Key techniques
- Region-based gesture detection
- Audio playback threading
- Interactive UI rendering
Allows hands-free browsing of image datasets using swipe navigation and pinch-to-zoom gestures.
Key techniques
- Gesture classification
- Coordinate transformation
- Dynamic image scaling
- Multi-panel UI composition
Transforms hand gestures into mouse movement, clicks, and drag operations for controlling web games.
Key techniques
- Cursor coordinate mapping
- Gesture threshold detection
- Motion smoothing filters
Implements YOLO for high-speed object detection on images and live webcam streams. Bounding boxes and labels are drawn in real time.
Key techniques
- Deep neural network inference
- Real-time object localization
- Video frame processing
Several core mathematical models are shared across projects.
Normalized landmark coordinates are mapped to pixel space: x = x_norm * W y = y_norm * H
x_c = sum(x * M) / sum(M) y_c = sum(y * M) / sum(M)
d = sqrt((x1 - x2)^2 + (y1 - y2)^2)
B_t = (1 - alpha) * B_(t-1) + alpha * I_t
Thresholding isolates significant regions for tracking and interaction.
pip install opencv-python mediapipe numpy pyautogui
- OpenCV installed and configured
- C++ compiler with OpenCV linkage
- Clone the repository.
- Navigate to the desired project folder.
- Ensure webcam access is enabled.
- Run the Python script or compile and execute the C++ program.
Each project folder contains independent source code and assets.
- Gesture-based interfaces
- Smart surveillance systems
- Interactive gaming
- Augmented reality visualization
- Educational demonstrations
- Human–computer interaction research
- Multi-hand gesture support
- GPU acceleration
- Deep learning gesture classification
- AR/VR integration
- Mobile deployment
Spriha Podder
Computer Vision Portfolio