datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
AIDLAB-HAR
AIDLAB-HAR
AIDLAB-HAR is a chest-worn inertial-sensor dataset for exercise recognition and repetition-marker detection. It contains synchronized recorder-frame acceleration and orientation quaternions sampled at 50 Hz.
The dataset accompanies Real-Time Sensor-Based Human Activity Recognition for eFitness and eHealth Platforms. This repository provides a corrected EDF/CSV distribution and viewer-friendly Parquet tables generated reproducibly from the immutable public source.… See the full description on the dataset page: https://huggingface.co/datasets/aidlab-wearables/AIDLAB-HAR.wearables_benchmarks
Wearables Benchmarks
This page consolidates the GitHub and HuggingFace links to wearables AI benchmarks from Meta.
Datasets highly relevant to wearable use cases
Benchmark
Paper
GitHub
HuggingFace
TLDR
WearVox
arXiv
facebookresearch/wearvox
WearVox
A benchmark for evaluating voice assistants in realistic wearable scenarios using 3,842 egocentric audio recordings collected via AI glasses across multiple task types like QA, tool calling, and speech translation.… See the full description on the dataset page: https://huggingface.co/datasets/kaimeta/wearables_benchmarks.