More class

[korosigSZTE]

Hi there,

What if I want to train more than one class? For example, 100 class, and each class has 500 images for training. It is impossible to load 50.000 images to memory. Or your model use generator to avoid this problem?

Do you have an example/solution for this?

[akTwelve]

Mask R-CNN can definitely handle that. The original researchers trained it with 90 classes and many more images.
https://github.com/matterport/Mask_RCNN

[korosigSZTE]

Thank you for your help....

Actually, I want to create a model which solve a multiclass classification problem.
The main concept is:

• every picture contain only one object
• the background is very simple
• all object is coming from the same object class (exampleÉ knives, hats, shoes, etc.)
• the model will learn and predict the name of the object. (example: the model learn all type of knives, and when it get an image it will tell us the name of the knife)

To be clear her is an example, I have 50 types of knife, and the output of the model has to recognize the correct name of the knife. Knife name could be: Chef's Knife,Heavy Duty Utility Knife,Boning Knife, etc.

To solve this problem, I have started to use annotated, segmented (masked) images (Coocolike dataset) and MASK RCNN model.

As a first step, I got a prediction, but I really don't know if I'm on the right way.

My questions are:

• For this problem, Mask RCNN could be the solution, or it is impossible to recognize a tiny difference between two objects from the same class (example Chef's Knife, Heavy Duty Utility Knife)?
• Have you ever seen almost the same problem/solution, with github repo?

[akTwelve]

I completely lost track of this because I was on vacation, sorry about that. I don’t know if you actually need Mask R-CNN if the images only contain one object. Seems like overkill. Maybe something simpler like imagenet?