Multimedia Laboratory

Scene Specific Detector via Transfer Learning

Meng Wang, Xiaogang Wang


The performance of a single-image monocular pedestrian detector has achieved the state-of-the-art level. However, it has been observed that when applied to a specific scene, its performance may drop significantly due to mismatch between the source dataset used to train the detector and samples in the target scene.

To tackle this issue, we propose an automatically adaptive pedestrian detector that re-train itself from samples collected from the new traffic scene. As more target samples are included in the training set, the mismatch problem will eventually be solved.

New samples from the target scene are collected by setting an easy detection criterion. Their labels are estimated by a basket of factors, including motion, appearance and semantic region. The combination of these factors is the confidence score, a soft-label probability value.

We propose a unified objective function based on linear SVM. The new formulation has two-fold highlights. First, it incorporates the confidence score instead of traditional hard labels. Second, it jointly optimizes the slack penalty and the label propagation functions. Finally, a new scene specific model is obtained from the optmization.


  • Automatical adaptation. No manual labeling required.
  • Suprising performance increase.
  • Semantic region assisting sample selection. This was not explored before.
  • Unsupervised label estimation through propagation.


Datasets used in this work are available below:

  • MIT Traffic Dataset
  • CUHK Square Dataset
  • Code

    Liblinear Prior A Liblinear extension to solve soft-label SVMs.


    If you use our codes or dataset, it is recommended to cite the following papers:

  • Meng Wang, Wei Li and Xiaogang Wang. Transferring a Generic Pedestrian Detector Towards Specific Scenes. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition 2012. June 17 - 21, 2012, Providence, Rhode Island, USA.
  • PDF    Poster
  • Meng Wang and Xiaogang Wang. Automatic Adaptation of a Generic Pedestrian Detector to a Specific Traffic Scene. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition 2011. June 21 - 23, 2011, Colorado Springs, Colorado, USA.
  • PDF    Poster


    Left: Before Adaptation. Right: After Adaptation

    Share This Page

    © 2013 Multimedia Laboratory

    The Chinese University of Hong Kong 香港中文大学

    Site Credits