yolov8_n_mask-refine_syncbn_fast_8xb16-500e_coco.py

_base_ = './yolov8_s_mask-refine_syncbn_fast_8xb16-500e_coco.py'

# This config will refine bbox by mask while loading annotations and
# transforming after `YOLOv5RandomAffine`

deepen_factor = 0.33
widen_factor = 0.25

model = dict(
    backbone=dict(deepen_factor=deepen_factor, widen_factor=widen_factor),
    neck=dict(deepen_factor=deepen_factor, widen_factor=widen_factor),
    bbox_head=dict(head_module=dict(widen_factor=widen_factor)))