TDA2EVM5777: For Creating the TIDL image Classification uscase

We have TIDL file IO use-case for  image classification. It is written in C

Hi,

In the earlier days, the name image segmentation was typically used to represent a two class problem - for example to distinguish background from foreground. However, Image Segmentation has increased in complexity (with the help of Deep Learning) and is able to distinguish between pixels belonging to various object types. 

Yes - the purpose of semantic segmentation is to do image segmentation.

But image segmentation is also used is a more generic context (locate objects and boundaries lines, curves, etc.) - which semantic segmentation does not do. The word semantic segmentation as it is typically used merely assigns an object type to each pixel in the image. For segmenting the pixels corresponding to each object instance separately, the name instance segmentation is used. 

You may read up more here: https://en.wikipedia.org/wiki/Image_segmentation

Best regards,

Hi,