Have a look at deep-learning-based instance segmentation with MVTec HALCON and locate objects in an image.
In this tutorial, we are going through the HDevelop example dl_instance_segmentation_workflow and highlight important steps.
As output, you get a bounding boxes and pixel-precise masks for all detected instances. Since the workflow (and partly the output) is very similar to deep-learning-based object detection, we will mainly focus on the differences of object detection and instance segmentation. For this, we will go through the HDevelop example dl_instance_segmentation_workflow and highlight important steps.
0:00 Introduction to semantic segmentation
1:05 Introduction to the HDevelop example
1:20 Samples for training and model parameters
2:23 Training the instance segmentation model
3:11 Inference results
In this video, MVTec HALCON 21.11 is used.