Back to Projects

Computer Vision Projects

Comprehensive computer vision implementations for my Computer Vision course at Pitt

Overview

A collection of advanced computer vision projects showcasing a range of techniques for image analysis and pattern recognition. This includes three main components: (1) Bag of Words Classification for treating images as collections of visual features for effective categorization, (2) Hough Transform implementation for detecting geometric shapes in complex images through parametric space transformation, and (3) a custom Deep Learning framework with CNN architectures for image recognition tasks.

Technologies

Python OpenCV PyTorch TensorFlow NumPy

Key Features

  • Feature extraction and classification using BoW model
  • Geometric pattern detection with Hough Transform
  • CNN-based deep learning for complex image recognition
  • Edge detection and parameter space visualization
  • Transfer learning and data augmentation capabilities

Implementation

  • Developed visual feature extraction and vocabulary generation for BoW
  • Implemented parametric transformations for shape detection
  • Created neural network architecture with custom components
  • Built preprocessing pipelines for various image analysis tasks
  • Designed visualization tools for model interpretation and evaluation
  • Optimized algorithms for performance across different image complexities