SOnusingh Which is the best way you would suggest to do floor segmentation for an image? Basically, I have oak ffc3p , and would like to check if pre existing models in luxonis model zoo can work for diverse cases.
KlemenSkrlj Hi @SOnusingh , Unfortunately we currently don't have any pre-trained model that would have explicit class for "floor" in our Zoo. You can check out FastSAM-S which is outputing panoptic segmentation mask so floor will also be segmented. But there won't be any specific class assigned to it so you'll have to add some additional postprocessing to the app. The other option is to try YOLO-P which predicts road segmentation but not sure if that fits your usecase. If you know of a public pretrained model outside of our ZOO you can check out the Conversion documentation and export into device suitable format. The last option would be to train the model yourself. Her you can use either our LuxonisTrain library or we also have a nice integration with Ultralytics library. I hope these references help, Klemen
