VAGNet anticipates accidents in dashcam videos using global features from VideoMAE-V2 combined with transformers and graphs, reporting higher average precision and mean time-to-accident on four benchmarks while running more efficiently than prior methods.
Unobserved heterogeneity and the statistical analysis of highway accident data.Analytic methods in accident re- search, 11:1–16, 2016
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
VAGNet: Vision-based Accident Anticipation with Global Features
VAGNet anticipates accidents in dashcam videos using global features from VideoMAE-V2 combined with transformers and graphs, reporting higher average precision and mean time-to-accident on four benchmarks while running more efficiently than prior methods.