ST-VFM reprograms frozen vision foundation models with temporal-aware token adapters and cross-prompt coordination, and reports state-of-the-art results on ten spatio-temporal forecasting benchmarks.
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
CONDITIONAL 1representative citing papers
citing papers explorer
-
Reprogramming Vision Foundation Models for Spatio-Temporal Forecasting
ST-VFM reprograms frozen vision foundation models with temporal-aware token adapters and cross-prompt coordination, and reports state-of-the-art results on ten spatio-temporal forecasting benchmarks.