def fpn_neck_config(test_step_name):
"""Return the class containing the corresponding attributes according to
the fpn_test_step_names."""
s = 64
in_channels = [8, 16, 32, 64]
feat_sizes = [s // 2**i for i in range(4)] # [64, 32, 16, 8]
out_channels = 8
feats = [
torch.rand(1, in_channels[i], feat_sizes[i], feat_sizes[i])
for i in range(len(in_channels))
]
if (fpn_test_step_names[test_step_name] == 0):
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
add_extra_convs=True,
num_outs=5)
elif (fpn_test_step_names[test_step_name] == 1):
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
add_extra_convs=False,
num_outs=5)
elif (fpn_test_step_names[test_step_name] == 2):
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
add_extra_convs=True,
no_norm_on_lateral=False,
norm_cfg=dict(type='BN', requires_grad=True),
num_outs=5)
elif (fpn_test_step_names[test_step_name] == 3):
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
add_extra_convs=True,
upsample_cfg=dict(mode='bilinear', align_corners=True),
num_outs=5)
elif (fpn_test_step_names[test_step_name] == 4):
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
add_extra_convs=True,
upsample_cfg=dict(scale_factor=2),
num_outs=5)
elif (fpn_test_step_names[test_step_name] == 5):
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
add_extra_convs='on_input',
num_outs=5)
elif (fpn_test_step_names[test_step_name] == 6):
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
add_extra_convs='on_lateral',
num_outs=5)
elif (fpn_test_step_names[test_step_name] == 7):
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
add_extra_convs='on_output',
num_outs=5)
return fpn_model, feats
def test_fpn():
"""Tests fpn."""
s = 64
in_channels = [8, 16, 32, 64]
feat_sizes = [s // 2**i for i in range(4)] # [64, 32, 16, 8]
out_channels = 8
# `num_outs` is not equal to len(in_channels) - start_level
with pytest.raises(AssertionError):
FPN(in_channels=in_channels,
out_channels=out_channels,
start_level=1,
num_outs=2)
# `end_level` is larger than len(in_channels) - 1
with pytest.raises(AssertionError):
FPN(in_channels=in_channels,
out_channels=out_channels,
start_level=1,
end_level=4,
num_outs=2)
# `num_outs` is not equal to end_level - start_level
with pytest.raises(AssertionError):
FPN(in_channels=in_channels,
out_channels=out_channels,
start_level=1,
end_level=3,
num_outs=1)
# Invalid `add_extra_convs` option
with pytest.raises(AssertionError):
FPN(in_channels=in_channels,
out_channels=out_channels,
start_level=1,
add_extra_convs='on_xxx',
num_outs=5)
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
start_level=1,
add_extra_convs=True,
num_outs=5)
# FPN expects a multiple levels of features per image
feats = [
torch.rand(1, in_channels[i], feat_sizes[i], feat_sizes[i])
for i in range(len(in_channels))
]
outs = fpn_model(feats)
assert fpn_model.add_extra_convs == 'on_input'
assert len(outs) == fpn_model.num_outs
for i in range(fpn_model.num_outs):
outs[i].shape[1] == out_channels
outs[i].shape[2] == outs[i].shape[3] == s // (2**i)
# Tests for fpn with no extra convs (pooling is used instead)
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
start_level=1,
add_extra_convs=False,
num_outs=5)
outs = fpn_model(feats)
assert len(outs) == fpn_model.num_outs
assert not fpn_model.add_extra_convs
for i in range(fpn_model.num_outs):
outs[i].shape[1] == out_channels
outs[i].shape[2] == outs[i].shape[3] == s // (2**i)
# Tests for fpn with lateral bns
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
start_level=1,
add_extra_convs=True,
no_norm_on_lateral=False,
norm_cfg=dict(type='BN', requires_grad=True),
num_outs=5)
outs = fpn_model(feats)
assert len(outs) == fpn_model.num_outs
assert fpn_model.add_extra_convs == 'on_input'
for i in range(fpn_model.num_outs):
outs[i].shape[1] == out_channels
outs[i].shape[2] == outs[i].shape[3] == s // (2**i)
bn_exist = False
for m in fpn_model.modules():
if isinstance(m, _BatchNorm):
bn_exist = True
assert bn_exist
# Bilinear upsample
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
start_level=1,
add_extra_convs=True,
upsample_cfg=dict(mode='bilinear', align_corners=True),
num_outs=5)
fpn_model(feats)
outs = fpn_model(feats)
assert len(outs) == fpn_model.num_outs
assert fpn_model.add_extra_convs == 'on_input'
for i in range(fpn_model.num_outs):
outs[i].shape[1] == out_channels
outs[i].shape[2] == outs[i].shape[3] == s // (2**i)
# Scale factor instead of fixed upsample size upsample
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
start_level=1,
add_extra_convs=True,
upsample_cfg=dict(scale_factor=2),
num_outs=5)
outs = fpn_model(feats)
assert len(outs) == fpn_model.num_outs
for i in range(fpn_model.num_outs):
outs[i].shape[1] == out_channels
outs[i].shape[2] == outs[i].shape[3] == s // (2**i)
# Extra convs source is 'inputs'
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
add_extra_convs='on_input',
start_level=1,
num_outs=5)
assert fpn_model.add_extra_convs == 'on_input'
outs = fpn_model(feats)
assert len(outs) == fpn_model.num_outs
for i in range(fpn_model.num_outs):
outs[i].shape[1] == out_channels
outs[i].shape[2] == outs[i].shape[3] == s // (2**i)
# Extra convs source is 'laterals'
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
add_extra_convs='on_lateral',
start_level=1,
num_outs=5)
assert fpn_model.add_extra_convs == 'on_lateral'
outs = fpn_model(feats)
assert len(outs) == fpn_model.num_outs
for i in range(fpn_model.num_outs):
outs[i].shape[1] == out_channels
outs[i].shape[2] == outs[i].shape[3] == s // (2**i)
# Extra convs source is 'outputs'
fpn_model = FPN(in_channels=in_channels,
out_channels=out_channels,
add_extra_convs='on_output',
start_level=1,
num_outs=5)
assert fpn_model.add_extra_convs == 'on_output'
outs = fpn_model(feats)
assert len(outs) == fpn_model.num_outs
for i in range(fpn_model.num_outs):
outs[i].shape[1] == out_channels
outs[i].shape[2] == outs[i].shape[3] == s // (2**i)