ZebraRecognizer: Pedestrian crossing recognition for people with visual impairment or blindness

作者：

Highlights：

• ZebraRecognizer identifies zebra crossings from a mobile device camera.

• The use of accelerometer and gyroscopes help remove projection distortion.

• The most expensive operations are run on the mobile device GPU.

• ZebraRecognizer is reliable: precision 1, recall .93 and avg distance error 0.2 m.

• The system is efficient: it can process about 25 frames per second on iPhone 5S.

摘要

Highlights•ZebraRecognizer identifies zebra crossings from a mobile device camera.•The use of accelerometer and gyroscopes help remove projection distortion.•The most expensive operations are run on the mobile device GPU.•ZebraRecognizer is reliable: precision 1, recall .93 and avg distance error 0.2 m.•The system is efficient: it can process about 25 frames per second on iPhone 5S.

论文关键词：Visual impairment,Blindness,Pedestrian crossing,Mobile computing

论文评审过程：Received 29 May 2015, Revised 9 November 2015, Accepted 9 May 2016, Available online 22 May 2016, Version of Record 15 June 2016.

论文官网地址：https://doi.org/10.1016/j.patcog.2016.05.002