一种加强SSD小目标检测能力的Atrous滤波器设计

针对实时目标检测SSD（single shot multiBox detector）算法对小目标检测能力偏差的问题，提出了一种提高特征图分辨率的Atrous滤波器设计策略。改进算法在SSD网络结构的基础上，把第三、四层卷积层产生的特征图经过规范化后连接在一起，然后通过Atrous卷积运算提高这些特征图分辨率。这些特征图共同提供小目标的所需的特征。另外该SSD改进算法还加入SeLU（scaled exponential linear units）激活函数，并在数据预处理阶段设计了一套数据增广方法。实验表明，该改进算法框架相对于原SSD算法框架具有更高的检测精度、更优良的鲁棒性，以及在小目标检测上效果明显。

Design of Atrous filter to strengthen small object detection capability of SSD

In order to overcome the shortcomings that SSD (Single Shot MultiBox Detector) can not detect small objects well, this paper proposed an Atrous filter design strategy, which can strengthen the resolution of feature maps. The improved algorithm concatenated the feature maps that generated by the third and fourth convolution layer after normalization, and then improves the resolution of these feature maps by Atrous computed. The concatenated feature maps provide the required features for small objects. In addition, the SSD improved algorithm also add SeLU (Scaled Exponential Linear Units) activation function and designed a data augmented methods in the data preprocessing phase. The experimental results shows that the proposal algorithm has higher detection accuracy and better robustness than the original SSD algorithm. Furthermore, the detection performance obvious better on small target detection.