Work

CREOLab: a procedure captioning dataset for understanding creative tool use in object-rich laboratory videos

Year: 2026

Type: article

Abstract: CREOLab is a laboratory video dataset for procedural captioning of human manual operations, designed to quantitatively evaluate object-knowledge biases in video captioning systems that may produce descriptions that differ from the actual procedures.

Source: Digital Discovery

DOI: 10.1039/d5dd00542f

Authors: Shigeaki Goto, Tatsuki Hasebe

Institution: Toyota Central Research and Development Laboratories (Japan)

Language: English

Topic: Multimodal Machine Learning Applications

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: diamond

OpenAlex ID: W7160868428

FWCI:

Cites: 32

Cited by:

Related to: