HoGs are a feature descriptor that has been widely and successfully used for object detection. It presents objects as a single feature vector as oppossed to a set of feature vectors where each represents a segment of the image. It’s computed by sliding window detector over an image, where a HOG descriptor is a computed for each position. Like SIFT the scale of the image is adjusted (pyramiding). HOGs are often used with SVM (Support Vector Machine) classifiers. Each HOG descriptor that is computed is fed to a SVM classifier t

HOG is a dense feature extraction method for images. Dense means that it extracts features for all locations in the image (or a region of interest in the image) as opposed to only the local neighborhood of keypoints like SIFT.

Histograms of Oriented Gradients features are widely used in computer vision applications for pattern extraction from images. HOG features are robust and easy to compute. HOG features compute the local object appearance and shape within an image using the distribution of intensity gradients or edge directions.HOG method for feature extraction is similar to Scale-Invariant Feature Transform (SIFT), edge oriented histograms, and shape contexts, but it is computed on a dense grid of uniformly spaced cells.

Here is an example of HoG with SVM:

 

Reference

HOGs for Human Detection by Dalal and Triggs