Hi! I'm using the instance segmentation augmentor on a coco format dataset. The package seems to use a huge amount of memory (up to 32gb) on images with many of annotations.
The following is my configuration:
PROBLEM = "instance_segmentation"
ANNOTATION_MODE = "coco"
INPUT_PATH = "..."
GENERATION_MODE = "linear"
OUTPUT_MODE = "coco"
OUTPUT_PATH = "..."
I would greatly appreciate any help / insights you can provide 👍