Why is the mAP only 0.43 when training the DIOR dataset with the same ImageNet backbone?

[likingliu]

Thank you for your work,but i have a question.
data = dict(
samples_per_gpu=8,
workers_per_gpu=1,
train=dict(pipeline=train_pipeline, version=angle_version),
val=dict(version=angle_version),
test=dict(version=angle_version))

optimizer = dict(
delete=True,
type='AdamW',
# lr=0.0001, #/8*gpu_number,
lr = 0.0001,
betas=(0.9, 0.999),
weight_decay=0.05)

[likingliu]

我也是