ASFF- 自适应空间特征融合论文及代码

https://github.com/ruinmessi/ASFF

文章：https://arxiv.org/pdf/1911.09516v2.pdf

1070 416 50ms （ 20fps）

1 增强的 baseline

为了更好地证明我们提出的ASFF方法的有效性，我们基于这些先进技术建立了一个比origin yolov3强得多的基线。

在高级训练技巧[43]( Bag of freebies for training object detection neural networks)之后，我们介绍了训练过程中的一些技巧，
如混合算法[12]、余弦[26]学习速率调度和同步批处理规范化技术[30]。除了这些技巧之外，我们还像[45]一样添加了一个无锚分支来与基于锚的分支一起运行，并利用[38]提出的锚引导机制来改进结果。此外，为了更好地进行包围盒回归，在原光滑L1损失上使用了额外的交并（IoU）损失函数[41]。

训练技巧

[43]( Bag of freebies for training object detection neural networks)

[12] Zhang Hongyi, Cisse Moustapha, N. Dauphin Yann, and David Lopez-Paz. mixup: Beyond empirical risk minimization. ICLR, 2018.

[26] Ilya Loshchilov and Frank Hutter. Sgdr: Stochastic gradient descent with warm restarts. arXiv preprint arXiv:1608.03983,2016.

[30] Chao Peng, Tete Xiao, Zeming Li, Yuning Jiang, Xiangyu Zhang, Kai Jia, Gang Yu, and Jian Sun. Megdet: A large mini-batch object detector. In CVPR, 2018

无锚点

[38] Jiaqi Wang, Kai Chen, Shuo Yang, Chen Change Loy, and
Dahua Lin. Region proposal by guided anchoring. In CVPR,
2019.

[45] Chenchen Zhu, Yihui He, and Marios Savvides. Feature selective anchor-free module for single-shot object detection.In CVPR, 2019.

bbox损失函数

Jiahui Yu, Yuning Jiang, Zhangyang Wang, Zhimin Cao, and Thomas Huang. Unitbox: An advanced object detection network. In ACMM, 2016.

利用上述先进技术，在COCO[22]2017 val集合上以50 FPS的（在特斯拉V100上）获得38.8%的mAP，大大提高了原始的YOLOv3-608基线（33.0%的mAP，52 FPS[31]），而无需大量的计算成本。

2 改进特征融合

不同于以往的基于元素和或级联的多层次特征融合方法，核心思想是自适应地学习各尺度特征地图融合的空间权重。如图2所示，它由两个步骤组成：相同的重新缩放和自适应融合。

尺度调整

x1 x2 x3 具有不同的分辨率和不同的通道数,去、为了融合要一致。

因此我们相应地修改了每个尺度的上采样和下采样策略。对于上采样，我们首先使用1×1卷积层将特征的通道数压缩到l级，然后分别使用插值来提高分辨率。对于1/2比例的下采样，我们简单地使用一个3×3的卷积层（步长为2）来同时修改通道数和分辨率。对于1/4的比例，我们在2步卷积之前添加了一个2步最大池层。

特征融合

alpha beta gama是0~1权重，并和为一。用三者加权。

为训练的参数

各层三部分可视化和融合后结果

梯度计算

链式法则+近似




对应融合系数
所以：

结果

class ASFF(nn.Module): 
     def __init__(self, level, rfb=False, vis=False): 
        super(ASFF, self).__init__() 
        self.level = level 
        self.dim = [512, 256, 256] 
        self.inter_dim = self.dim[self.level] 
        # 每个level融合前，需要先调整到一样的尺度
        if level==0: 
            self.stride_level_1 = add_conv(256, self.inter_dim, 3, 2) 
            self.stride_level_2 = add_conv(256, self.inter_dim, 3, 2) 
            self.expand = add_conv(self.inter_dim, 1024, 3, 1) 
        elif level==1: 
            self.compress_level_0 = add_conv(512, self.inter_dim, 1, 1) 
            self.stride_level_2 = add_conv(256, self.inter_dim, 3, 2) 
           self.expand = add_conv(self.inter_dim, 512, 3, 1) 
       elif level==2: 
           self.compress_level_0 = add_conv(512, self.inter_dim, 1, 1) 
           self.expand = add_conv(self.inter_dim, 256, 3, 1) 
       compress_c = 8 if rfb else 16  #when adding rfb, we use half number of channels to save memory 

       self.weight_level_0 = add_conv(self.inter_dim, compress_c, 1, 1) 
       self.weight_level_1 = add_conv(self.inter_dim, compress_c, 1, 1) 
       self.weight_level_2 = add_conv(self.inter_dim, compress_c, 1, 1) 

       self.weight_levels = nn.Conv2d(compress_c*3, 3, kernel_size=1, stride=1, padding=0) 
       self.vis= vis 
       
    def forward(self, x_level_0, x_level_1, x_level_2): 
        if self.level==0: 
           level_0_resized = x_level_0 
           level_1_resized = self.stride_level_1(x_level_1) 
 
           level_2_downsampled_inter =F.max_pool2d(x_level_2, 3, stride=2, padding=1) 
           level_2_resized = self.stride_level_2(level_2_downsampled_inter) 
 
       elif self.level==1: 
           level_0_compressed = self.compress_level_0(x_level_0) 
           level_0_resized =F.interpolate(level_0_compressed, scale_factor=2, mode='nearest') 
           level_1_resized =x_level_1 
           level_2_resized =self.stride_level_2(x_level_2) 
       elif self.level==2: 
           level_0_compressed = self.compress_level_0(x_level_0) 
           level_0_resized =F.interpolate(level_0_compressed, scale_factor=4, mode='nearest') 
           level_1_resized =F.interpolate(x_level_1, scale_factor=2, mode='nearest') 
          level_2_resized =x_level_2 
 
       level_0_weight_v = self.weight_level_0(level_0_resized) 
       level_1_weight_v = self.weight_level_1(level_1_resized) 
       level_2_weight_v = self.weight_level_2(level_2_resized) 
       levels_weight_v = torch.cat((level_0_weight_v, level_1_weight_v, level_2_weight_v),1) 
       # 学习的3个尺度权重
       levels_weight = self.weight_levels(levels_weight_v) 
       levels_weight = F.softmax(levels_weight, dim=1) 
       # 自适应权重融合
       fused_out_reduced = level_0_resized * levels_weight[:,0:1,:,:]+\ 
                           level_1_resized * levels_weight[:,1:2,:,:]+\ 
                           level_2_resized * levels_weight[:,2:,:,:] 
 
       out = self.expand(fused_out_reduced) 
 
       if self.vis: 
           return out, levels_weight, fused_out_reduced.sum(dim=1) 
       else: 
          return out 

ubuntu环境

安装apex

https://github.com/NVIDIA/apex

Compile the DCN layer
./make.sh

更新torchvision
torchvision’ has no attribute ‘ops’

https://pypi.org/project/torchvision/#files
下载whl文件安装