FalconApp converts iPhone video captures of rigid objects into on-device perception models for masks and 6-DoF poses using automatically labeled synthetic data from GSplat reconstruction, with 20-minute generation time and better accuracy than PnP on 4/5 tested objects.
Bdd100k: A diverse driving dataset for heterogeneous multitask learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
FalconApp: Rapid iPhone Deployment of End-to-End Perception via Automatically Labeled Synthetic Data
FalconApp converts iPhone video captures of rigid objects into on-device perception models for masks and 6-DoF poses using automatically labeled synthetic data from GSplat reconstruction, with 20-minute generation time and better accuracy than PnP on 4/5 tested objects.