22 lines
513 B
Python
22 lines
513 B
Python
model = dict(
|
|
type='ImageClassifier',
|
|
backbone=dict(
|
|
type='ViTEVA02',
|
|
arch='b',
|
|
img_size=224,
|
|
patch_size=14,
|
|
sub_ln=True,
|
|
final_norm=False,
|
|
out_type='avg_featmap'),
|
|
neck=None,
|
|
head=None,
|
|
)
|
|
|
|
data_preprocessor = dict(
|
|
# RGB format normalization parameters
|
|
mean=[0.48145466 * 255, 0.4578275 * 255, 0.40821073 * 255],
|
|
std=[0.26862954 * 255, 0.26130258 * 255, 0.27577711 * 255],
|
|
# convert image from BGR to RGB
|
|
to_rgb=True,
|
|
)
|