Need to understand autoaugment in the case of object detection on large images

[ramdhan1989]

anyone can help me to make me understand cfg file below. Since there is a random crop transformation, do I need to feed images in original size? or do I need to do image slicing outside the module? I would like to use my custom dataset using this repo :

https://github.com/fcakyon/small-object-detection-benchmark/blob/main/mmdet_configs/visdrone_tood/tood_crop_480_960_cls_60.py

I am not sure if random crop here is used to slice images (including the associated bounding boxes) or it is just for augmentation.

# dataset settings
img_norm_cfg = dict(mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
train_pipeline = [
    dict(type="LoadImageFromFile"),
    dict(type="LoadAnnotations", with_bbox=True),
    dict(
        type="AutoAugment",
        policies=[
            [
                dict(type="RandomCrop", crop_type="absolute_range", crop_size=(480, 640), allow_negative_crop=True),
                dict(type="Resize", img_scale=(1333, 800), keep_ratio=True),
            ],
            [
                dict(type="RandomCrop", crop_type="absolute_range", crop_size=(480, 640), allow_negative_crop=True),
                dict(type="Resize", img_scale=(1333, 800), keep_ratio=True),
            ],
            [
                dict(type="Resize", img_scale=(1333, 800), keep_ratio=True),
            ],
        ],
    ),
    dict(type="RandomFlip", flip_ratio=0.5),
    dict(type="Normalize", **img_norm_cfg),
    dict(type="Pad", size_divisor=32),
    dict(type="DefaultFormatBundle"),
    dict(type="Collect", keys=["img", "gt_bboxes", "gt_labels"]),
]

Thank you

[ccomkhj]

slice images (including the associated bounding boxes)