This is the repository for MMASD: A Multimodal Dataset for Autism Intervention Analysis. Dataset can be accessed at https://tinyurl.com/mr3rd2dp.
Theme 1 -- Robotic-assisted therapy: Arm Swing, Body Swing, Chest Expansion, Squat
Theme 2 -- Rhythm: Drumming, Maracas Forward Shaking, Maracas Shaking, Sing and Clap
Theme 3 -- Yoga: Frog Pose, Tree Pose, Twist Pose
We presents a novel privacy-preserving open-source dataset, MMASD as a Multi Modal ASD benchmark dataset, collected from play therapy interventions of children with Autism.
MMASD includes data from 32 children with ASD, and 1,315 data samples segmented from over 100 hours of intervention recordings.
It covers 11 different activities in 3 themes, and consists of four privacy-preserving modalities of data:
Optical flow: derived from raw RGB video, for each frame, we have **_x.jpg and **_y.jpg, representing the horizontal and vertical components, respectively. Note originally optical flow is saved as .npy files, but due to the size limit, we convert them to .jpg format by mapping the optical flow to gray scale range.
Clinician ASD evaluation scores of children, including demographic information (age, gender, etc.) and autism related evaluation scores such as ADOS-2.
Due to the file size limit, we have uploaded the data to google drive, which can be easily accessed at https://tinyurl.com/mr3rd2dp.
Citation:
Jicheng Li, Vuthea Chheang, Pinar Kullu, Eli Brignac, Zhang Guo, Anjana Bhat, Kenneth E. Barner, and Roghayeh Leila Barmaki. MMASD: A Multimodal Dataset for Autism Intervention Analysis. In Proceedings of the 25th International Conference on Multimodal Interaction (ICMI '23), pp. 397โ405, 2023.