A new outdoor multiview multimodal QA dataset and a VAE-fused LVLM are presented with claims of large gains over prior models, but the evaluation is clouded by likely train-test source overlap and an undefined average.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
City-VLM: Towards Multidomain Perception Scene Understanding via Multimodal Incomplete Learning
A new outdoor multiview multimodal QA dataset and a VAE-fused LVLM are presented with claims of large gains over prior models, but the evaluation is clouded by likely train-test source overlap and an undefined average.