Sign in to view source links and access this dataset
Description
OctoNet is a multi-modal dataset of human activity recordings from multiple sensor modalities, including inertial measurement units (IMU), motion capture, and mmWave/Radar data. The dataset is hosted by the author 'hku-aiot' on Hugging Face and was last updated on 2025-05-16. It is intended for research in activity recognition, pose estimation, and multi-modal data fusion.
Use Cases
Train human activity recognition models based on the described IMU data.
Develop pose estimation algorithms based on the provided motion capture data.
Research multi-modal data fusion techniques using the combined IMU, motion capture, and radar data streams.
Strengths
Data includes multiple sensor modalities, which likely provides complementary information for analysis.
Motion capture data is provided in both CSV and .npy formats, offering flexibility for different processing pipelines.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count is unknown, which may limit suitability assessment.
Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
Source
hku-aiot
Collection Method
Recordings of human activities from multiple sensor modalities.
Freshness
Last updated 2025-05-16 04:08:51; freshness should be verified.
License is unknown; users must verify terms of use before downloading.