基于候选区域和并行卷积神经网络的行人检测

针对行人在部分自然场景图像中所占比例较小（以下简称小目标），提取的特征容易丢失，检测准确率低的问题，提出基于候选区域和并行卷积神经网络（Parallel Convolutional Neural Network，PCNN）的行人检测方法。对于候选区域提取部分，改进了选择性搜索，使其更符合行人这一类别的候选区域提取；利用Edge Boxes对选择性搜索提取的大量预候选区域进行过滤，最终得到数量少、质量高的候选区域。在利用卷积神经网络（Convolutional Neural Network，CNN）进行特征提取时，针对深层卷积神经网络能够提取到更丰富更抽象的高层特征，但同时对于小目标容易造成特征丢失的问题，加入浅层网络组成并行卷积神经网络（Parallel Convolutional Neural Network，PCNN）提取深、浅层特征输出。最后将所提方法应用于行人检测，实验结果表明，所提方法对于小目标的检测准确率有较好的提升。

Pedestrian Detection Based on Candidate Regions and Parallel Convolutional Neural Network

Aiming at the problems that the percentage of pedestrians in some natural scenes is small（hereinafter referred to as small target）, the extracted features are easily lost, and the detection accuracy is low, a pedestrian detection method based on candidate regions and Parallel Convolutional Neural Network（PCNN） is proposed. First, for the candidate region extraction section, selective search is improved to make it more suitable for pedestrians in this category of candidate region extraction. Then, Edge Boxes are used to filter a large number of pre-candidate regions extracted by selective search. Finally, a small number of high-quality candidate regions are obtained. When using Convolutional Neural Network（CNN） to extract features, deeper convolutional neural networks can extract richer and more abstract high-level features, but at the same time, the small objects can easily cause feature loss, adding shallow convolutional neural network to build a parallel convolutional neural network to extract deep and shallow feature outputs. Finally, the proposed method is applied to pedestrian detection. The experimental results show that the proposed method can improve the detection accuracy of small target.