Transfer Learning Classification with MMPretrain

[christian-holland]

Please correct me if I am missing something, but as far as I understood training "custom" datasets by using the given configs does not use any pretrained weights (e.g. for Resnet18), as this repo is meant to do the actual pretraining.

Now if I instead do want to do some classic finetuning for classification tasks, using pretrained weights, is there some recommendation how to approach this task with this repo?

[Ezra-Yu]

how to ft refer to this tutorial.

as the example:

_base_ = [
    '../_base_/models/resnet50.py',           # model settings
    '../_base_/datasets/imagenet_bs32.py',    # data settings
    '../_base_/schedules/imagenet_bs256.py',  # schedule settings
    '../_base_/default_runtime.py',           # runtime settings
]

# >>>>>>>>>>>>>>> Override model settings here >>>>>>>>>>>>>>>>>>>
model = dict(
    backbone=dict(
        frozen_stages=2,
        init_cfg=dict(
            type='Pretrained',
            checkpoint='https://download.openmmlab.com/mmclassification/v0/resnet/resnet50_8xb32_in1k_20210831-ea4938fc.pth',
            prefix='backbone',
        )),
    head=dict(num_classes=10),
)
# <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<\

model.backbone.init_cfg = dict(type='Pretrained', checkpoint="path_or_url", prefix='backbone',) do the load pretrained weight configuration.