Prior Knowledge Guided Small Object Detection on High-Resolution Images

Zixuan Yang, Xiujuan Chai, Ruiping Wang, Weijun Guo, Weixuan Wang, Li Pu, Xilin ChenICIP 2019 · OpenReview

Abstract

When applying common object detection algorithms to detect small objects on high-resolution images, the down-sampling operation of the input images is inevitable due to the limitation of GPU memory. Accordingly, the details for characterizing small objects are lost. To resolve this contradiction, a small object detection method in a coarse-to-fine manner is presented. Specifically, some rough regions of interest (ROI) are firstly computed from low-resolution images. The prior knowledge of the positions of objects is used to guide the generation of ROIs. Then the features of small ROIs are recomputed from high-resolution images, and the features of large ROIs are obtained from the feature maps used to generate ROIs. The proposed method is validated on two datasets. One is a plant phenotyping dataset and the other is a public traffic sign dataset. Experimental results convincingly show the effectiveness of the proposed method.
BibTeX
@inproceedings{yang2019prior,
  author = {Zixuan Yang and Xiujuan Chai and Ruiping Wang and Weijun Guo and Weixuan Wang and Li Pu and Xilin Chen},
  title = {{Prior Knowledge Guided Small Object Detection on High-Resolution Images}},
  booktitle = {ICIP},
  year = {2019},
  url = {https://openreview.net/forum?id=pHk8-ur0bGH},
}

Loading related content...

Figure