Text/Human Detection

Scale-based Region Growing For Scene Text Detection

  • Patent Application Number: 201210241520.2
  • A paper of this work is published in ACM Multimedia 2013 as a full paper. [pdf]
  • The MSRG scene text database: [download]
  • In this work, we present a novel scale-based region growing algorithm for scene text detection. We first distinguish SIFT features in text regions from those in background by exploring the inter- and intra-statistics of SIFT features. Then scene text regions in images are identified by scale-based region growing, which explores the geometric context of SIFT keypoints in local regions. Our algorithm is very effective to detect multilingual text in various fonts, sizes, and with complex background. In addition, it offers insights on efficiently deploying local features in numerous applications, such as visual search. We evaluate our algorithm on three datasets including a newly constructed multilingual dataset, and achieve the state-of-the-art performance.

Campus management based on the analysis of the videos

  • Research Program for USTC-Microsoft Research Asia (MSRA): Smart Campus
  • We aim to build a campus management system based on the analysis of the content of a monitoring video. We have finished a part of the system to report the crowdedness of the refectory, playground and study room according to our survey on students’ demands. The system possessed two cooperated algorithms. One algorithm counts the number of the people entering and leaving the place. Its accumulation fault will be rectified by another algorithm, which directly calculates the crowdedness. This service is available on a website. As a part of this program, we also built an acoustic control robot. I took responsibility of programming the speech recognition chip (Lingyang SPCE061A).
      Click the thumbnails to see the big images