🚶🏻‍➡🚶🏻‍♀️‍➡️ Pedestrian Image Classification

🎯 Objective : Binary image classification of images into pedestrian vs. non-pedestrian using traditional feature extraction and machine learning techniques

• Individual component of the course project for the Computer Vision (COMP9517) course during the year 2020 at UNSW Sydney
• Achieved best performance using Histogram of Oriented Gradients (HOG) with 1 cell per block for feature extraction and Support Vector Machine (SVM) for classification.
• Reached 0.999 AUC and 0.989 accuracy on held-out test data

More details about the approach and score are available in the project report.

---

Contents

• Project Report
• Main Code
• Code for Generating Report Figures
• Evaluation Results
• Data Files