User-Centered Design for Custom Workflow Configuration in Video Analytics Systems: A Usability and McCall Quality Evaluation

Abstrak

Artificial intelligence (AI)-based video analytics systems increasingly rely on configurable workflows to manage data processing and model inference; however, many existing visual workflow and low-code approaches remain difficult for operational users due to complex interfaces and technical terminology. This study investigates how a User-Centered Design (UCD) approach can improve the usability of custom workflow configuration in video analytics environments. A node-based workflow interface was developed following a UCD process that includes context-of-use analysis, user requirement specification, iterative design, and evaluation. The usability of the proposed interface was assessed through a formative usability study using the System Usability Scale (SUS), complemented by observational findings from task-based evaluations, whereas system behavior was examined through structured functional testing interpreted using selected attributes of the McCall quality model, focusing on usability, reliability, and correctness. The evaluation involved operational users of an industrial video analytics platform performing representative workflow configuration tasks, resulting in an average SUS score of 64, indicating acceptable but improvable usability. Further analysis shows that the proposed design supports better workflow comprehension, reduces configuration errors during task execution, and assists users in verifying workflow correctness through validation and preview mechanisms. This study contributes an evaluative perspective that connects user-centered interface design with both perceived usability and functional system behavior in configurable AI workflow systems.

Identifier

• OpenAlex: https://openalex.org/W7163755378
• DOI: https://doi.org/10.48084/etasr.18601