一张图理解Faster—RCNN测试流程

请教同学Faster—RCNN测试流程,他让我看一个Faster—RCNN模型文件(自己找哈):一张图理解Faster—RCNN测试流程_第1张图片




看了后,自己画了一张图:

注:conv:卷积              relu:(rectified linear units)激活函数          fc:全连接             RPN:region proposal network 

cls_prob:分类概率       bbox_pred:bounding box predit

...:代表 relu norm pool (为了美观不画)


一张图理解Faster—RCNN测试流程_第2张图片

如果你看过Faster—RCNN论文,一定会想起这张图:



一张图理解Faster—RCNN测试流程_第3张图片

Figure 2: Faster R-CNN is a single, unified network
for object detection. The RPN module serves as the
‘attention’ of this unified network.


不要问我为什么放倒了,和上面的图对应,有木有????(为了美观我把右下的箭头指向了右上!)

图上的单词我还用红笔标注了。


论文提到ZF共享的五个卷积层不就是左边conv1-conv5吗?

再看文章里这段:To generate region proposals, we slide a small network over the convolutional feature map output by the last shared convolutional layer. This small network takes as input an n * n spatial window of the input convolutional feature map. Each sliding window is mapped to a lower-dimensional feature (256-d for ZF and 512-d for VGG, with ReLU [33]following). This feature is fed into two sibling fully connected layers—a box-regression layer (reg) and a box-classification layer (cls).

 last shared convolutional layer 不就是conv5吗?reg cls 不就对应 fc7后面的两个分支吗?


顿时思路清晰了,感觉自己棒棒哒。。。。。。


你可能感兴趣的:(深度学习,一张图)