Improving the performance of pedestrian detectors using convolutional learning

作者：

Highlights：

• This paper proposes a cascade of non-deep detectors with a CNN model.

• Pedestrian Detection is addressed by testing on INRIA and Caltech datasets.

• Different non-deep detectors are used: ACF, LDCF and Spatial Pooling + (SP+).

• State-of-the-art competitive performance is achieved cascading SP+ with the CNN.

摘要

Highlights•This paper proposes a cascade of non-deep detectors with a CNN model.•Pedestrian Detection is addressed by testing on INRIA and Caltech datasets.•Different non-deep detectors are used: ACF, LDCF and Spatial Pooling + (SP+).•State-of-the-art competitive performance is achieved cascading SP+ with the CNN.

论文关键词：Pedestrian detection,Convolutional neural network,Feature maps,Non-deep detectors

论文评审过程：Received 29 January 2016, Revised 19 May 2016, Accepted 24 May 2016, Available online 27 May 2016, Version of Record 13 October 2016.

论文官网地址：https://doi.org/10.1016/j.patcog.2016.05.027