@@ -373,7 +373,7 @@ class DPN(nn.Module):
self.last_linear = nn.Conv2d(in_chs, num_classes, kernel_size=1, bias=True)
def logits(self, features):
- if not self.training and self.test_time_pool:
+ if not self.training and self.test_time_pool and False:
x = F.avg_pool2d(features, kernel_size=7, stride=1)
out = self.last_linear(x)
# The extra test time pool should be pooling an img_size//32 - 6 size patch
@@ -423,8 +423,9 @@ def adaptive_avgmax_pool2d(x, pool_type='avg', padding=0, count_include_pad=Fals
else:
if pool_type != 'avg':
print('Invalid pool type %s specified. Defaulting to average pooling.' % pool_type)
- x = F.avg_pool2d(
- x, kernel_size=(x.size(2), x.size(3)), padding=padding, count_include_pad=count_include_pad)
+ #x = F.avg_pool2d(
+ # x, kernel_size=(x.size(2), x.size(3)), padding=padding, count_include_pad=count_include_pad)
+ x = F.adaptive_avg_pool2d(x, (1, 1))
return x
@@ -459,4 +460,4 @@ class AdaptiveAvgMaxPool2d(torch.nn.Module):
def __repr__(self):
return self.__class__.__name__ + ' (' \
+ 'output_size=' + str(self.output_size) \
- + ', pool_type=' + self.pool_type + ')'
\ No newline at end of file
+ + ', pool_type=' + self.pool_type + ')'