Developing an instance segmentation application
Like object detection, instance segmentation also involves object localization and classification. However, instance segmentation takes one step further while localizing the detected objects of interest. Specifically, besides classification, models for this task localize the detected objects at the pixel level. The following sections detail the steps to develop an instance segmentation application using Detectron2 pre-trained models.
Selecting a configuration file
Like object detection, Detectron2 also provides a list of cutting-edge models pre-trained for object instance segmentation tasks. For instance, Figure 2.5 shows the list of Mask R-CNN models pre-trained on the COCO Instance Segmentation dataset.
Figure 2.5: COCO Instance Segmentation baselines with Mask R-CNN
After checking the specifications of these models, this specific application selects the X101-FPN model. The following is the relative...
