A transformer-based high-level fusion method (HiLO) improves F1 by up to 25.9 points over Kalman-filter baselines on a real urban and highway dataset, with code and data promised publicly.
”Simpletrack: Understanding and rethinking 3d multi- object tracking.” ECCV , 2022
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
HiLO: High-Level Object Fusion for Autonomous Driving using Transformers
A transformer-based high-level fusion method (HiLO) improves F1 by up to 25.9 points over Kalman-filter baselines on a real urban and highway dataset, with code and data promised publicly.