CapProbe converts detailed caption evaluation into region-aligned multiple-choice checking with 25,650 questions and finds large coverage gaps and a competency-efficiency trade-off across 13 VLMs.
Introducing Qwen3.7-Plus.https: //qwen.ai/blog?id=qwen3.7-plus, 2026
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CapProbe: Evaluating Detailed Image Captions via Full-Scene Dense Question Answering
CapProbe converts detailed caption evaluation into region-aligned multiple-choice checking with 25,650 questions and finds large coverage gaps and a competency-efficiency trade-off across 13 VLMs.