A wearable AI system that turns the live view into one sentence and back into an image lets users experientially confront how linguistic mediation filters and biases perception.
Transparent Human Evaluation for Image Captioning
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
We establish THumB, a rubric-based human evaluation protocol for image captioning models. Our scoring rubrics and their definitions are carefully developed based on machine- and human-generated captions on the MSCOCO dataset. Each caption is evaluated along two main dimensions in a tradeoff (precision and recall) as well as other aspects that measure the text quality (fluency, conciseness, and inclusive language). Our evaluations demonstrate several critical problems of the current evaluation practice. Human-generated captions show substantially higher quality than machine-generated ones, especially in coverage of salient information (i.e., recall), while most automatic metrics say the opposite. Our rubric-based results reveal that CLIPScore, a recent metric that uses image features, better correlates with human judgments than conventional text-only metrics because it is more sensitive to recall. We hope that this work will promote a more transparent evaluation protocol for image captioning and its automatic metrics.
citation-role summary
citation-polarity summary
fields
cs.HC 1years
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
SEMANTIC SEE-THROUGH GOGGLES: Wearing Linguistic Virtual Reality in (Artificial) Intelligence
A wearable AI system that turns the live view into one sentence and back into an image lets users experientially confront how linguistic mediation filters and biases perception.