ScholarWorks@성균관대학교

초록

Sketch-based drawing assessments in art therapy are commonly used to understand the cognitive and psychological states of individuals. In conjunction with self-report measures, drawing assessments serve to enhance insights into an individual's psychological state. However, interpreting the drawing assessments is labor-intensive and substantially reliant on the experience of the art therapists. While a few automated approaches for analyzing drawing-based assessments have been proposed to remedy this issue, they mostly rely on existing object detection methods, where complex drawing attributes cannot be accurately decoded. To overcome these challenges, we propose a novel and comprehensive Draw-A-Person-in-the-Rain (DAPR) analysis system, CheckDAPR, which utilizes a Multimodal Large Language Model (MLLM) with object detection methods for in-depth evaluation. Our experimental results show the promising performance of CheckDAPR and its ability to reduce analysis time for art therapists, indicating its potential to aid professionals in art therapy.

키워드