The goal of the project is to experiment with different features for classifying vehicle vs non-vehicle. Then I used the training to detect vehicles on the road.
- get_hog_features() function in helper.py from line to .
I started by reading in one car and one not car image, called get_hog_features to get features and hog images.
Below is the result comparing side by side:
I then explored different color space and different skimage.hog() parameters (orientations, pixels_per_cell, and cells_per_block). I grabbed random images from each of the two classes and displayed then to get a feel for what the _skimages.hog() output look like.
Here is an example using the YCrCb color space and HOG parameters of orientation=9, pixels_per_cell=(8,8) and cell_per_block=(2,2)
- I tried various combinations of parameters and I decided to use YCrCb color space with HOG parameters of
orientations=9,pixels_per_cell=(8, 8)andcells_per_block=(2, 2)because it shows the different features output between car and not car.