Reproduce “COTTA: Context-Aware Transfer Adaptation for Trajectory Prediction in Autonomous Driving” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Seohyoung Park, Jaeyeol Lim, Seoyoung Ju, Kyeonghun Kim, Nam-Joon Kim, Hyuk-Jae Lee
The paper evaluates four strategies for adapting QCNet pretrained on Argoverse 2 to South Korean vehicle trajectories. Encoder freezing keeps the pretrained encoder fixed and fine-tunes only the decoder on ETRI data. It reports minADE of 0.182, versus 0.540 from scratch, a 66.3% reduction. This result is limited to one dataset and experiment.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.