Challenge: Abnormal Activity Detection in Individuals with Developmental Disabilities

Submission Dates:
02/01/2025 to 02/28/2025
Citation Author(s):
Taihei
Fujioka
Kyushu Institute of Technology
Christina
Garcia
Kyushu Institute of Technology
Sozo
Inoue
Kyushu Institute of Technology
Submitted by:
Christina Garcia
Last updated:
Mon, 03/17/2025 - 02:22
DOI:
10.21227/qfkw-sa40
Data Format:
License:
Creative Commons Attribution

Abstract 

Facilities for the developmentally disabled face the challenge of detecting abnormal behaviors because of limited staff and the difficulty of spotting subtle movements. Traditional methods often struggle to identify these behaviors because abnormal actions are irregular and unpredictable, leading to frequent misses or misclassifications.

Currently, many facilities for persons with disabilities rely on direct observation and manual recording processes, which increase staff workload and make it difficult to quickly and accurately identify abnormal behaviors. However, manual recording has limitations, making it challenging to continuously collect sufficient data.

The main contributions of this work is the creation of a unique dataset with labeled abnormal behaviors. The dataset includes labeled video data collected for four days from five normal participants performing eight activities with four abnormal behaviors. The data was collected with normal participants to simulate activities, and no individuals with disabilities.

For evaluation, we assessed all normal versus abnormal activities and per abnormal activity recognition comparing with the baseline. The dataset was collected in a laboratory-controlled environment with a total of five participants in a span of four days recorded at a sampling rate of 30 frames per second. In this study, participants' actions were recorded using two different cameras, and pose estimation techniques were used to extract skeleton data.

The selection of these activities was based on observations and interviews conducted in a facility for individuals with developmental disabilities. Facility staff identified these behaviors as frequently occurring and requiring attention to ensure the safety and well-being of individuals. The abnormal activities were chosen to reflect patterns of self-injurious or aggressive behavior commonly observed in the facility, including "head banging", "throwing objects", "attacking others", and "biting hands or fingers". Similarly, the normal activities represent routine daily actions observed in the same environment, such as "sitting," "walking," "using a phone," and "eating or drinking." 

Instructions: 

1. The zip contains data of 5 participants with label "keypoints_with_labels_P(user_id)"
2. Each csv file contains the keypoints extracted with activity labels on the last column.
3. The user_id for each csv matched the user_id of the file name.