Robot House Multiview Human Activity Recognition Dataset

Brief Description

RHM-HAR-1 is a new multi-View RGB dataset of human activities including multiple viewpoints. Its main characteristics are:

• Four perspectives: two side views, one mobile robot view, one top (ceiling) view
• 14 activities of daily living (classes): Walking, Sitting Down, Standing Up, Lifting Objects, Carrying Objects, Drinking, Stairs Climbing Up, Stairs Climbing Down, Stretching, Putting Objects Down, Reaching, Opening Can, Closing Can, and Cleaning.
• 407-700 recordings for each class (from four perspectives)
• 6,701 total recordings (from four viewpoints), in 26,804 videos

The data is split into training (65%), testing (20%), and validation (15%) for each view. All video files are named after the following schema: ClassName_PerspectiveName_ClipNumber.avi. Clips with the same class name and number with various view names are synchronised.

Access

Please contact robothouse@herts.ac.uk about how to access the dataset RHM-HAR-1.

• Part 0
• Part 1
• Part 2
• Part 3
• Part 4
• Part 5
• Part 6
• Part 7
• Part 8
• Part 9
• Part 10
• Part 11
• Part 12
• Part 13
• Part 14
• Part 15
• Part 16
• Part 17
• Part 18
• Part 19
• Part 20
• Part 21
• Part 22
• Part 23

Referencing

Please use the following format for referencing the dataset in a scientific context. The paper can be downloaded free of charge at the ThinkMind Digital Library.

Mohammad Hossein Bamorovat Abadi, Mohamad Reza Shahabian Alashti, Patrick Holthaus, Catherine Menon, and Farshid Amirabdollahian.
RHM: Robot House Multi-view Human Activity Recognition Dataset.
In The Sixteenth International Conference on Advances in Computer-Human Interactions (ACHI 2023).
Venice, Italy, in press. IARIA.