Hi there! I have an upcoming project, wherein I am planning to create a model for object detection by its size (in this case a dragon fruit for example). To be specific, the model will be able to identify the dragon fruit whether it is small, medium, or large. Will instance segmentation work best here in this case?
Instance segmentation could work! If the camera is in a fixed location, you can use bounding boxes or instance segmentation. In either case, you’ll be converting pixels to a distance. These two posts may help:
Thanks for the response! I am considering using instance segmentation, but I notice that object detection can also be used. I read the articles here that object detection is faster and generally produces more accurate results than the instance segmentation, which the latter requires a bigger-sized dataset. In this instance, is using object detection also okay? I am anticipating that I will be working on a limited-size data.
You can always try both and see which yields better results!
1 Like
This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.