Keras-具有两个类Tensorflow Python的图像分类

Question

i want to classify two classes with CNN.

• first class are objects(car, building, and so go an)
• secound class are NoObjects(for example Background without objects)

So, the question is, how can I realize this? I want to have many objects images for the CNN. I saw a example with Cat and Dog.. But I have only images for one class. The other class are the background. It is possible to create a background class??

I´m very new in CNN..

Answer 1

It seems that you're talking about semantic segmentation , where you have a variable N number of classes to localize (in this case N=1) plus the background class. There's a lot of literature and baseline models out there that can help you out. I recommend these two since they're the most popular ones:

• Pretrained models: Tensorflow's model zoo
• Datasets: Specifically COCO