MHAD: Multimodal Home Activity Dataset
The MHAD dataset was jointly collected by JD Health, Huazhong University of Science and Technology, and Zhejiang University. It is the first multimodal dataset captured in real home environments, featuring multiple camera angles and a wide range of household scenarios. It includes the most comprehensive set of physiological signals to date and is a valuable resource for computer vision, machine learning, and biomedical engineering research.
Dataset description and usage context
MHAD: Multimodal Home Activity Dataset with Multi‑Angle Videos and Synchronized Physiological Signals
Description
MHAD is a multimodal home‑activity dataset collected jointly by JD Health, Huazhong University of Science and Technology, and Zhejiang University. It comprises multi‑angle videos and synchronized physiological signals captured in authentic home settings, making it the first publicly available dataset of its kind. The dataset includes extensive physiological recordings, providing a rich resource for computer‑vision, machine‑learning, and biomedical‑engineering research.
Content
- Physiological Signals: Heart rate, respiration rate, and other key biosignals.
- Video Data: Multi‑angle video recordings covering diverse household scenes.
- Annotations: Detailed labels for each scene and corresponding physiological signals.
Access and Use
- Purpose: Academic research only; commercial use is prohibited.
- Access Procedure:
- Send a request email from an official academic address to feijintao3@jd.com.
- Use the subject line “MHAD Dataset Access Request”.
- Include in the body:
- Full name
- Academic institution
- Position
- Brief research description and intended use of the dataset
- Confirmation that the dataset will be used solely for academic purposes
- Await review (typically 5‑7 business days).
- Upon approval, receive an email containing the download link.
Dataset Structure
MHAD_Dataset
-----------------
MHAD/
|-- sub01/
| |-- a/ # Before exercise
| | |-- 1/ # Watching TV
| | | |-- output1.avi # frontal view
| | | |-- output2.avi # 90‑degree side view
| | | |-- output3.avi # 45‑degree side view
| | | |-- sub01_a_1.csv # ground‑truth
| | |-- 2/ # Using phone
| | | |-- output1.avi
| | | |-- output2.avi
| | | |-- output3.avi
| | | |-- sub01_a_2.csv
| | |-- 3/ # Reading
| | | |-- output1.avi
| | | |-- output2.avi
| | | |-- output3.avi
| | | |-- sub01_a_3.csv
| | |-- 4/ # Talking
| | | |-- output1.avi
| | | |-- output2.avi
| | | |-- output3.avi
| | | |-- sub01_a_4.csv
| | |-- 5/ # Eating
| | | |-- output1.avi
| | | |-- output2.avi
| | | |-- output3.avi
| | | |-- sub01_a_5.csv
| | |-- 6/ # Drinking
| | | |-- output1.avi
| | | |-- output2.avi
| | | |-- output3.avi
| | | |-- sub01_a_6.csv
| |-- b/ # After exercise
| | |-- 1/
| | | |-- output1.avi
| | | |-- output2.avi
| | | |-- output3.avi
| | | |-- sub01_b_1.csv
| | |-- 2/
| | | |-- output1.avi
| | | |-- output2.avi
| | | |-- output3.avi
| | | |-- sub01_b_2.csv
| | |-- 3/
| | | |-- output1.avi
| | | |-- output2.avi
| | | |-- output3.avi
| | | |-- sub01_b_3.csv
| | |-- 4/
| | | |-- output1.avi
| | | |-- output2.avi
| | | |-- output3.avi
| | | |-- sub01_b_4.csv
| | |-- 5/
| | | |-- output1.avi
| | | |-- output2.avi
| | | |-- output3.avi
| | | |-- sub01_b_5.csv
| | |-- 6/
| | | |-- output1.avi
| | | |-- output2.avi
| | | |-- output3.avi
| | | |-- sub01_b_6.csv
|-- sub40/
| ...
-----------------
Pair the dataset with AI analysis and content workflows.
Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.