Search

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 could 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 improved the resolution of these feature maps by Atrous computed. The concatenated feature maps provided the required features for small objects. In addition, the SSD improved algorithm also added 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. [ABSTRACT FROM AUTHOR]

In order to overcome the shortcomings that low detection accuracy and slow network training of the real-time object detection algorithm YOLO, this paper proposed an algorithm that added the batch re-normalization to the YOLO network structure. It combined the characteristic of batch re-normalization, which had the strength to deal with small or non-i. i. d. minibatches. The improved algorithm treated the feature maps, which generated from the convolutional layers, as activations, and then batched renormalize the activations, meanwhile it removed the dropout from the original network structure and increased learning rate. The experimental results show that the proposal algorithm has better detection accuracy and faster than before, and furthermore, it can decrease the model training time and the requirement of hardware equipment. [ABSTRACT FROM AUTHOR]