I currently have a problem that I want to find a specific template in the image. The goal that I would like to get from the tensor flow is [X, y, angle]. x and y are the coordinates of the location of the point inside the template. the angle is the angle of rotation of the image on the image 0-359 ° (in radius).
Also, the template does not change mainly, but depending on where it happens, because the images are created from a 2D laser scanner. Therefore, sometimes you sometimes see parts of it.
The input is a 400x400 black and white image, where white is the point of the laser scanner and black is the free space.
What is the problem? Classification? But how do I get the position where it is found in the image? In fact, I think this is a bit like the one Google uses to find things in images.
I really modified the Deep MNIST example, but somehow I didn't get to the point of getting something useful from it.
Does anyone have an idea that could point me in the right direction?
Hi,
Christian
source
share