InstanceCap generates instance-level structured video captions by isolating objects with auxiliary models and prompting MLLMs with class-specific hints, improving text-to-video generation fidelity when used for finetuning.
Survey of hallucination in natural language generation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
InstanceCap: Improving Text-to-Video Generation via Instance-aware Structured Caption
InstanceCap generates instance-level structured video captions by isolating objects with auxiliary models and prompting MLLMs with class-specific hints, improving text-to-video generation fidelity when used for finetuning.