person detection using HOG features with SVM

Hi, please I need help in person detection. I have positive and negative images to use in training SVM classifier. I read about using HOG features descriptor and I downloaded some HOG features matlab programs, the input for these programs are the image vector, the window size, the cell size, number of bin, and the overlapping value. I don't know how to calculate these variables (except the image vector)and the effect of changing them, and if the HOG program is used to extract the features for both the training images and the test image Please if any one have experience in this field explain and advise me. the second question, after calculating the HOG to each image and used in training the SVM. In detection process we have the test image and the training classifier. what is the proper method used to detect if the object is in the image or not????? Thank you to every one can help

[Breezy Boo]

https://www.mathworks.com/help/vision/ref/vision.peopledetector-class.html