Hello everyone! I’ve been tinkering with Lightship lately and it looks very cool especially for its semantic segmentation. Is it possible for us to add our custom segment? I’m thinking to detect whether we can detect cats or dogs.
Cheers!
Hi Giri,
Currently available semantic channels are the sky, ground, artificial ground, water, buildings, and foliage (https://lightship.dev/docs/semantics.html#autotoc_md76), but we do not offer other categories as you mentioned. Detection of new categories of environmental objects and surface types (e.g., people, animals, flowers, etc.) are in the development phase but they aren’t slated for any particular release, so they don’t appear on any roadmap at this time.
For the most current information on available categories, please see our roadmap.,
https://storage.googleapis.com/nianticweb-media/lightship/lightship-roadmap.png
Thank you for your patience.
1 Like
I also would love to see the ability to create custom segments. There are so many powerful ideas it would open up.