mirrors/pytorch-image-models
1
0
Fork 0
mirror of https://github.com/huggingface/pytorch-image-models.git synced 2025-06-03 15:01:08 +08:00
Code Issues Projects Releases Wiki Activity

Stricter check on pool_type for adaptive pooling module. Fix #2159

Browse Source
This commit is contained in:
Ross Wightman 2024-05-03 16:16:51 -07:00
parent d6da4fb01e
commit 301d0bb21f
1 changed files with 7 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
timm/layers/adaptive_avgmax_pool.py
View File

@ -134,6 +134,7 @@ class SelectAdaptivePool2d(nn.Module):
super(SelectAdaptivePool2d, self).__init__() super(SelectAdaptivePool2d, self).__init__()
assert input_fmt in ('NCHW', 'NHWC') assert input_fmt in ('NCHW', 'NHWC')
self.pool_type = pool_type or '' # convert other falsy values to empty string for consistent TS typing self.pool_type = pool_type or '' # convert other falsy values to empty string for consistent TS typing
pool_type = pool_type.lower()
if not pool_type: if not pool_type:
self.pool = nn.Identity() # pass through self.pool = nn.Identity() # pass through
self.flatten = nn.Flatten(1) if flatten else nn.Identity() self.flatten = nn.Flatten(1) if flatten else nn.Identity()
@ -145,8 +146,10 @@ class SelectAdaptivePool2d(nn.Module):
self.pool = FastAdaptiveAvgMaxPool(flatten, input_fmt=input_fmt) self.pool = FastAdaptiveAvgMaxPool(flatten, input_fmt=input_fmt)
elif pool_type.endswith('max'): elif pool_type.endswith('max'):
self.pool = FastAdaptiveMaxPool(flatten, input_fmt=input_fmt) self.pool = FastAdaptiveMaxPool(flatten, input_fmt=input_fmt)
else: elif pool_type == 'fast' or pool_type.endswith('avg'):
self.pool = FastAdaptiveAvgPool(flatten, input_fmt=input_fmt) self.pool = FastAdaptiveAvgPool(flatten, input_fmt=input_fmt)
else:
assert False, 'Invalid pool type: %s' % pool_type
self.flatten = nn.Identity() self.flatten = nn.Identity()
else: else:
assert input_fmt == 'NCHW' assert input_fmt == 'NCHW'
@ -156,8 +159,10 @@ class SelectAdaptivePool2d(nn.Module):
self.pool = AdaptiveCatAvgMaxPool2d(output_size) self.pool = AdaptiveCatAvgMaxPool2d(output_size)
elif pool_type == 'max': elif pool_type == 'max':
self.pool = nn.AdaptiveMaxPool2d(output_size) self.pool = nn.AdaptiveMaxPool2d(output_size)
else: elif pool_type == 'avg':
self.pool = nn.AdaptiveAvgPool2d(output_size) self.pool = nn.AdaptiveAvgPool2d(output_size)
else:
assert False, 'Invalid pool type: %s' % pool_type
self.flatten = nn.Flatten(1) if flatten else nn.Identity() self.flatten = nn.Flatten(1) if flatten else nn.Identity()
def is_identity(self): def is_identity(self):