I'm trying to build a weed detection model using IBM Visual Recognition service. I have successfully annotated my images with bounding box but the when i try to evaluate my model on a new test image, the model misclassified the new image. I want to know if IBM WATSON support other labelling technique aside from Bounding Box. technique like polygon annotation, semantic, line e-t-c
No, it only supports bounding boxes for object detection.