/Person-Detection-SVM-HOG

Person detector using Histograms of Oriented Gradients (HOGs) for feature extraction, and a linear SVM for classification. Done for my Honours ML project. Includes image data I collected.

Primary LanguageJupyter Notebook

Person-Detection-SVM-HOG

Person detector using Histograms of Oriented Gradients (HOGs) for feature extraction, and a linear SVM for classification. Done for my Honours Adaptive Computation and Machine Learning project. Includes image data I collected.

How to run the demo (requires webcam)

Option 1: (may take some time)

  • Run all cells except the ones in the Predictions and performance on test set section.
  • Run Demo cell

Option 2: (only necessary cells)

  • Run the first 3 cells under Loading images with corresponding labels & extracting features using HOG.
  • Run both cells under Splitting data into training & testing sets.
  • Run the cell containing the Linear_SVM class cell and the cell under Fit SVM on training data
  • Run Demo cell

Requirements

Python packages

  • numpy
  • cv2
  • matplotlib
  • scikit-image
  • cvxopt
  • pandas
  • scikit-learn
  • bokeh