[solved] – Question 87431

In this HW, you are asked to implement histogram of oriented gradient (HOG) descriptor from scratch as discussed in class. HOG is not just used in interest point descriptors such as SIFT. It is also a widely used feature for object recognition.

For simplicity, we will ignore the step of finding interest points but simply partition the image into cell/patch and compute the HOG in each cell.

You can use Matlab. (Note that the Matlab demo appears to display the gradient with 90o rotation and thus shows something a bit different.) “

p/s: I NEED the source code and picture result of the (file-20170526-6421-1j3azw3.jpg) image. I uploaded my professor’s example images and his resulting picture (gantrycrane.png and hog_result.png) to show you the result I want for the ((file-20170526-6421-1j3azw3.jpg). You can cut half of the picture if it’s too big.

Expert Answer

---