Computer Vision & Machine Learning Portfolio A collection of end-to-end Machine Learning and Computer Vision projects focusing on real-time detection, object localization, and image classification.
Projects Overview:-
- Real-Time Landmark Detection Tech: Python, MediaPipe, OpenCV, NumPy
Description: Implemented a high-performance landmark detection system capable of tracking facial contours, hand gestures, and body pose in real-time.
Key Features: Uses MediaPipe’s holistic solutions to map coordinates for 468+ facial landmarks and hand skeletons.
Use Case: Can be extended for sign language recognition or driver drowsiness detection.
- Object Detection with TensorFlow Tech: TensorFlow, TF Hub, Matplotlib, OpenCV
Description: Developed an object detection pipeline using pre-trained deep learning models (SSD/MobileNet) to identify and localize multiple objects within a single frame.
Key Features: Implemented non-max suppression and bounding box visualization. The model is optimized for mobile-friendly inference.
Use Case: Surveillance systems or automated inventory tracking.
- Multi-Class Pet Classification Tech: TensorFlow, Keras, Pandas, Matplotlib
Description: Built a Convolutional Neural Network (CNN) to classify various breeds of pets using the Oxford-IIIT Pet Dataset.
Key Features: Applied image preprocessing, data augmentation, and categorical cross-entropy loss to achieve high classification accuracy.
Use Case: Animal healthcare apps or automated pet identification systems.
Tech Stack:- Languages: Python
Deep Learning: TensorFlow, Keras
Computer Vision: OpenCV, MediaPipe
Data Science: NumPy, Pandas, Matplotlib
Environment: VS Code, Google Colab