越线人群计数--Crossing-line Crowd Counting with Two-phase Deep Neural Networks
越线人群计数--Crossing-line Crowd Counting with Two-phase Deep Neural Networks
Crossing-line Crowd Counting with Two-phase Deep Neural Networks ECCV2016
人群计数有两种做法：1） region-of-interest (ROI) 感兴趣区域计数，2） line-of-interest (LOI) 感兴趣线计数，就是通过图像中某一条线的人流量。
大多数算法关注 ROI 计数问题，但是我们认为 LOI 计数问题更有实际应用价值，对于人群密集的大型公共区域，我们想通过监控相机对所有区域进行人群计数是不现实的。但是我们可以对出口和入口进行 LOI 计数从而推理出公共区域里的人数总量。
传统的 LOI 方法是通过提取视频中 对应 LOI 那条线上的信息生成 2D temporal slices generating 2D temporal slices by temporally concatenating video frame lines at the LOI
基于 temporal slices 我们来估计通过画线的人流总量，这种方法在某些情况下效果不太好 When the scene is not crowded and pedestrians walks in normal speed (Fig. 1(b, row 1)), people can be well recognized in the temporal slices. However, we observe that temporal slices are not robust to scenes with high crowd density, slow walking speed, and low camera viewing angles. In Fig. 1(b, row2), the temporal slice shows excess jitters and people in it are no longer recognizable
本文使用 CNN 网络从视频序列中学习特征，分两个步骤来解决计数问题
3 Method 3.1 Pixel-level supervision maps
根据视频序列帧，我们得到中间结果 Density map 和 Velocity map，有了这两个信息，再去得到 Supervision crowd counting map The crowd counting map can then be obtained as the elementwise multiplication of the density and velocity maps 相应位置数值相乘得到 crowd counting map
3.2 Deep Convolutional Neural Network for LOI crowd counting
我们的网络结构如下：
We propose to train the CNN in two phases. In the first training phase, the network is trained to predict the crowd density map D t and the crowd velocity map V t simultaneously 我们是分步训练上面的网络的，首先训练 predict the crowd density map 和 crowd velocity map， 再微调 fine-tune the trained network in the second phase with supervision of the crossing-line crowd counting map
3.3 From crowd counting map to LOI counts 这一步主要是根据 crowd counting map 我们计算得到画线的随时间变化的人流量
4 Experiments 我们针对越线人群计数，建立了一个新的数据库
