How do you adjust the bounding box during image augmentation?

vision
1

When image augmentation is applied to input images, how do you adjust the labels of the bounding box? I’m trying to implement data augmentation to a YOLO model.

2

You could try to reuse some transformations from detection/transforms.py or use an approach from e.g. this topic.

1 Like
3

imgaug or albumentations may help you.

Some Examples:
imgaug: Examples: Bounding Boxes
albumentations: Bounding boxes augmentation

1 Like