ASPP: Atrous Spatial Pyramid Pooling
deeplabv2中的aspp如上图所示,在特征顶部映射图使用了四中不同采样率的空洞卷积。这表明以不同尺度采样时有效的,在Deeolabv3中向ASPP中添加了BN层。不同采样率的空洞卷积可以有效捕获多尺度信息,但会发现随着采样率的增加,滤波器有效权重(权重有效的应用在特征区域,而不是填充0)逐渐变小。如下图
当我们以不同采样率的3*3卷积核应用在65*65的特征映射上,当采样率接近特征映射大小时,3*3滤波器不是捕获全图像上下文,而是退化为简单的1*1滤波器,只有滤波器中心起作用。
为了克服这个问题,我们考虑使用图片级特征。具体来说,我们在模型最后的特征映射上应用全局平均,将结果经过1x1的卷积,再双线性上采样得到所需的空间维度。最终我们改进的ASPP包括:
-
1.一个1x1卷积和三个3x3的采样率为rates={6,12,18}的空洞卷积,滤波器数量为256,包含BN层。针对output_stride=16的情况。如下图(a)部分Atrous Spatial Pyramid Pooling
-
2.图像级特征,即将特征做全局平均池化,经过卷积再融合。如下图(b)部分Image Pooling
也就是说当output_stride=8时,加倍了采样率。所有特征通过1x1级联到一起,生成最终的分数。代码(pytorch)为:
class ASPP(nn.Module):
def __init__(self, in_channel=512, depth=256):
super(ASPP,self).__init__()
# global average pooling : init nn.AdaptiveAvgPool2d ;also forward torch.mean(,,keep_dim=True)
self.mean = nn.AdaptiveAvgPool2d((1, 1))
self.conv = nn.Conv2d(in_channel, depth, 1, 1)
# k=1 s=1 no pad
self.atrous_block1 = nn.Conv2d(in_channel, depth, 1, 1)
self.atrous_block6 = nn.Conv2d(in_channel, depth, 3, 1, padding=6, dilation=6)
self.atrous_block12 = nn.Conv2d(in_channel, depth, 3, 1, padding=12, dilation=12)
self.atrous_block18 = nn.Conv2d(in_channel, depth, 3, 1, padding=18, dilation=18)
self.conv_1x1_output = nn.Conv2d(depth * 5, depth, 1, 1)
def forward(self, x):
size = x.shape[2:]
image_features = self.mean(x)
image_features = self.conv(image_features)
image_features = F.upsample(image_features, size=size, mode='bilinear')
atrous_block1 = self.atrous_block1(x)
atrous_block6 = self.atrous_block6(x)
atrous_block12 = self.atrous_block12(x)
atrous_block18 = self.atrous_block18(x)
net = self.conv_1x1_output(torch.cat([image_features, atrous_block1, atrous_block6,
atrous_block12, atrous_block18], dim=1))
return net
本文内容参考: