mirror of
https://github.com/open-mmlab/mmselfsup.git
synced 2025-06-03 14:59:38 +08:00
65 lines
1.7 KiB
Python
65 lines
1.7 KiB
Python
# Copyright (c) OpenMMLab. All rights reserved.
|
|
import copy
|
|
import platform
|
|
|
|
import pytest
|
|
import torch
|
|
|
|
from mmselfsup.data import SelfSupDataSample
|
|
from mmselfsup.models.algorithms import MoCo
|
|
|
|
queue_len = 32
|
|
feat_dim = 2
|
|
momentum = 0.999
|
|
backbone = dict(
|
|
type='ResNet',
|
|
depth=18,
|
|
in_channels=3,
|
|
out_indices=[4], # 0: conv-1, x: stage-x
|
|
norm_cfg=dict(type='BN'))
|
|
neck = dict(
|
|
type='MoCoV2Neck',
|
|
in_channels=512,
|
|
hid_channels=2,
|
|
out_channels=2,
|
|
with_avg_pool=True)
|
|
head = dict(
|
|
type='ContrastiveHead',
|
|
loss=dict(type='mmcls.CrossEntropyLoss'),
|
|
temperature=0.2)
|
|
|
|
|
|
@pytest.mark.skipif(platform.system() == 'Windows', reason='Windows mem limit')
|
|
def test_moco():
|
|
data_preprocessor = {
|
|
'mean': (123.675, 116.28, 103.53),
|
|
'std': (58.395, 57.12, 57.375),
|
|
'bgr_to_rgb': True
|
|
}
|
|
|
|
alg = MoCo(
|
|
backbone=backbone,
|
|
neck=neck,
|
|
head=head,
|
|
queue_len=queue_len,
|
|
feat_dim=feat_dim,
|
|
momentum=momentum,
|
|
data_preprocessor=copy.deepcopy(data_preprocessor))
|
|
assert alg.queue.size() == torch.Size([feat_dim, queue_len])
|
|
|
|
fake_data = [{
|
|
'inputs': [torch.randn((3, 224, 224)),
|
|
torch.randn((3, 224, 224))],
|
|
'data_sample':
|
|
SelfSupDataSample()
|
|
} for _ in range(2)]
|
|
|
|
fake_inputs, fake_data_samples = alg.data_preprocessor(fake_data)
|
|
fake_loss = alg(fake_inputs, fake_data_samples, mode='loss')
|
|
assert fake_loss['loss'] > 0
|
|
assert alg.queue_ptr.item() == 2
|
|
|
|
# test extract
|
|
fake_feats = alg(fake_inputs, fake_data_samples, mode='tensor')
|
|
assert fake_feats[0].size() == torch.Size([2, 512, 7, 7])
|