Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
OBJECT DETECTION METHOD AND APPARATUS
Document Type and Number:
WIPO Patent Application WO/2019/148362
Kind Code:
A1
Abstract:
Disclosed are an object detection method and apparatus. The method comprises: extracting a global image feature from an input image by means of a plurality of convolutional layers; determining a plurality of candidate detection regions by means of the global image feature; according to the information and by means of a predetermined number of convolutional layers in the plurality of convolutional layers, extracting a first local image feature corresponding to the predetermined number of convolutional layers; determining a second local image feature of each candidate detection region in the plurality of candidate detection regions according to the first local image feature, wherein second local image features of some candidate detection regions in the plurality of candidate detection regions are determined by means of extracted first local image features corresponding to at least two convolutional layers in the predetermined number of convolutional layers; and performing object detection according to the second local image feature of each candidate detection region, and outputting an object detection result. Therefore, the balance between a spatial resolution and semantic information can be realized during the extraction of a local image feature, thereby improving the detection accuracy of an object.

Inventors:
BAI XIANGHUI (CN)
TAN ZHIMING (CN)
Application Number:
PCT/CN2018/074706
Publication Date:
August 08, 2019
Filing Date:
January 31, 2018
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
FUJITSU LTD (JP)
BAI XIANGHUI (CN)
TAN ZHIMING (CN)
International Classes:
G06K9/62; G06K9/46; H04N5/232
Domestic Patent References:
WO2017173605A12017-10-12
Foreign References:
CN107463892A2017-12-12
CN107341517A2017-11-10
Attorney, Agent or Firm:
BEIJING SANYOU INTELLECTUAL PROPERTY AGENCY LTD. (CN)
Download PDF: