I am currently working on a project in which i'm supposed to classify whether an image contains a translucent watermark or not. This is hard to do with standard object classification or template matching because of the translucency and very low image quality.

I thought about generating a dataset with the watermark pasted somewhere on the image. Given enough generated data, could I train a neural network to classify the real images? If not, why not? Thank you for your help.

  • $\begingroup$ You may want to use spatial transformers, but you may need a lot of data. $\endgroup$ – Media Nov 24 '18 at 17:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.