Resources: Datasets & Software

Some of the lectures and practical tutorials will use existing, third-party datasets. In addition, students will be given instructions on how to collect data from the sensors and use some of the interactive and robotic technology in the testbed, as part of the prototype they will build during the 3-days school.

Datasets used in human activity recognition lectures:

Tim Van Kasteren's dataset: https://sites.google.com/site/tim0306/datasets

CASAS datas: http://casas.wsu.edu/datasets/

Reccomended software for activity recognition: Matlab

Recommended software for object detection:

If you plan to perform object detection (YOLOv3) on your computer during the summer school, you should install python3 and the latest version of OpenCV for Python.

On a Linux/Unix machine:

python3 -m venv yolo-env

source yolo-env/bin/activate

pip3 install numpy

pip3 install opencv-python

On Windows, this page is helpful:

https://www.scivision.dev/install-opencv-python-windows/

Docker images could be an option for Macs, but have not been tested:

https://www.learnopencv.com/install-opencv-docker-image-ubuntu-macos-windows/

Further instructions will be provided at the summer school.

Software to program the humanoid robot Pepper from Softbank Robotics:

See at http://doc.aldebaran.com/2-5/dev/community_software.html how to download the software,

You will need to create a SoftBank Robotics Community user account (https://community.ald.softbankrobotics.com/en/resources/software/language/en-gb)

and download Pepper Software Suite 2.5.5 for your OS.

Smart Home

Instructions on how to access the sensors and actuators in the smart home will be provided during the school.

Students are recommended to use the OpenHAB's REST interface: https://www.openhab.org/docs/configuration/restdocs.html