2D Instance Segmentation Configuration Guide
About 4066 wordsAbout 14 min
This guide explains what the parameters do and how to tune them during the 2D instance segmentation stage of recognition. First, configure the 2D instance segmentation parameters shared by all relevant tasks. If you use the Loading and unloading of plane type target objects (Target Independent) task, continue with the instance segmentation 3D parameters specific to that task.
2D Instance Segmentation
The following parameters adjust the instance detection results produced by the deep learning model. Start with the default values, then adjust parameters one at a time based on missed detections, false detections, bounding box coverage, and confidence scores.
Related Documents
Optimize the input image and depth data before instance segmentation
Select the instances passed to downstream processes
Set the order in which instances are processed
Generate the point cloud needed for subsequent 3D calculations