WEEK4 RESEARCH Amari Lewis Aidean Sharghi
PREPARING THE DATASET Cars – 83 samples 3 images for each sample when x=0 7 images for each sample when y=0 Buildings- 80 samples 3 images for each sample when x=0 7 images for each sample when y=0 1630
RUNNING EPI PROGRAM Extracting the EPI for all categories When y=0; And when x=0; The primary category is Cars and Buildings, the other EPIs from the other categories: Trees, Buses, and Bikes will be used as negative data (classification).
RUNNING HOG Dense- Histogram of Oriented Gradients – type of feature descriptor Extracted features from the EPI images to train a classifier *Due to the images sizes, takes very long time up to 5hours.
In order to compare our results, we ran HOG program on a.jpg image when x=0 and y=0 for each individual sample. Extracted features
NEXT STEP… Principal Component Analysis- run PCA Apply Fischer kernel Train classifier Test data