Experiment for real-time human behavior awareness by Feature And Body-part Learning (FABL)
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
mat
README.md
dataset.zip
drawConfusionMatrix.m
feature_extraction.m
main.m
train_test_generation.m

README.md

FABL

Experiment for real-time human behavior awareness by Feature And Body-part Learning (FABL) on real-world baxter interaction application, see http://hcr.mines.edu/code/FABL.html

Data format:

The dataset used in the experiment is compressed as dataset.zip. The data format of each .txt file is shown as follows:

's1_a2_e3.txt' stores the data when performing activity 2 by subject 1 for the third time. There are 5 columns in the data file

Frame # Skeletal joint # x coordinate y coordinate z coordinate
1 1 1.43 0.03 0.32
1 2 1.41 0.03 0.10
... ... ... ... ...
47 15 0.89 -0.36 -1.01

Usage:

  1. Uncompress dataset.zip to the project;

  2. Run 'feature_extraction.m' to extract features for all human activity data;

  3. Run 'train_test_generation.m' to split training and testing;

  4. Run 'main.m' to get the final results.

Cite:

If you use our method and/or codes, please cite our paper

@INPROCEEDINGS { han2017simultaneous,
      AUTHOR    = {Fei Han AND Xue Yang AND Christopher Reardon AND Yu Zhang AND Hao Zhang},
      TITLE     = {Simultaneous Feature and Body-Part Learning for Real-Time Robot Awareness of Human Behaviors},
      BOOKTITLE = {IEEE International Conference on Robotics and Automation (ICRA)},
      YEAR      = {2017}
  }