Would be nice to have Object Segmentation algorithm.
Thanks for the feature request! We'd love to hear more about how you would use this.
Basic explanation that I need to find similar objects on picture and get their shapes from picture and also knowing what is the object. Image processing with libraries like OpenCV does not produce enough quality and no information about found objects in case it found something.
I hope I'm not mixing with other algorithm, maybe it's Semantic Segmentation. :)
@sakrist Thanks! Knowing that there is a demand for a semantic segmentation toolkit is good to know during feature planning.
Is it possible to add new toolkits ourselves? I would be interested in Semantic Segmentation as well.
@colegleason - Turi Create is 100% open source. So it is possible for anyone to add a new toolkit. However creating a new toolkit is a significant amount of work. I'd encourage you to get in contact with us early in the process. Perhaps opening up a "work in progress" code review once you have a stubbed out API you are considering using.
I am also interested in object segmentation. Specially to work with the vision framework.
Most helpful comment
@sakrist Thanks! Knowing that there is a demand for a semantic segmentation toolkit is good to know during feature planning.