This nodelet is used to compute the saliency of an image.

The image can be composed of any feature space. However, the current nodelet takes only 3 channel BGR image and computes the saliency map.

The original code is implemented by the authors of the paper: Human detection using a mobile platform and novel features derived from a visual saliency mechanism.

Please cite this paper should you use the code.

However, there are minor changes from the original code including the MultiThreading support using OpenMP.

Subscribing Topic

  • ~input (sensor_msgs/Image)

    Currently only supports BGR8 encoding (3 channel RGB intensity image)

Publishing Topic

  • /saliency_map_generator/output/saliency_map (sensor_msgs/Image)

    Output saliency map.


  • ~num_threads (Int, default: 2)

    Number of threads used for execution

  • ~fps (Bool, default: true)

    Prints the frame rate to the output image if true.


roslaunch jsk_perception sample_saliency_map_generator.launch