What model should I use to create some masks?

General Discussion
, ,
#1

Hi, I am new to TF and wanted to ask for ideas or approaches to my problem.

As a starting point, I have pictures of oesophagi and I would like to filter out the oesophagi in these. Using OpenCV and a refined revision with Paint, I have already created some masks (see pictures).

blue_channel366
blue_channel366968×716 87.6 KB

Result366
Result366968×716 2.76 KB

blue_channel747
blue_channel747692×512 302 KB

Result747
Result747692×512 2.4 KB

Unfortunately, I haven’t found a general algorithm for image processing so that I can get a good mask, so I wanted to try this with the help of TF. So far I have looked at the approach of an image segmentation, but this is rather used for several objects to be classified.

My question now would be:
Which approach would be the best for this problem and is there maybe a good guide for it?

Thanks in advance for any help!

#3

Hi @vincent_krause

Welcome to the TensorFlow Forum!

You can try using Tensorflow’s Semantic Segmentation with Model Garden by referring to the attached link. There are some other models also available such as Instance Segmentation or Object Detection model with Model Garden which you can use as per your requirement.