T
Trevor Darrell
Researcher at University of California, Berkeley
Publications - 734
Citations - 222973
Trevor Darrell is an academic researcher from University of California, Berkeley. The author has contributed to research in topics: Computer science & Object detection. The author has an hindex of 148, co-authored 678 publications receiving 181113 citations. Previous affiliations of Trevor Darrell include Massachusetts Institute of Technology & Boston University.
Papers
More filters
Proceedings ArticleDOI
Zero-shot Policy Learning with Spatial Temporal Reward Decomposition on Contingency-aware Observation
TL;DR: The authors decompose the sparse reward into particular regions in a contingency-aware observation as a per step reward, and then use Model Predictive Control (MPC) to obtain a policy.
Journal ArticleDOI
Does unsupervised grammar induction need pixels?
Boyi Li,Rodolfo Corona,Karttikeya Mangalam,Catherine S. Hsia Chen,Daniel Flaherty,S. Belongie,Kilian Q. Weinberger,Jitendra Malik,Trevor Darrell,Daniel Klein +9 more
TL;DR: This article investigated whether extralinguistic signals such as image pixels are crucial for inducing constituency grammars, and found that they persist in the presence of rich information from large language models.
Proceedings ArticleDOI
G^3: Geolocation via Guidebook Grounding
TL;DR: Lu et al. as mentioned in this paper proposed the task of Geolocation via Guidebook Grounding that uses a dataset of StreetView images from a diverse set of locations and an associated textual guidebook for GeoGuessr, a popular interactive geolocation game.
Journal Article
Explaining Reinforcement Learning Policies through Counterfactual Trajectories
Julius Frost,Olivia Watkins,Eric Weiner,Pieter Abbeel,Trevor Darrell,Bryan A. Plummer,Ksenia Saenko +6 more
TL;DR: This work generates trajectories that conveys how the agent performs under distribution shifts by showing the agent’s behavior across a wider trajectory distribution, and demonstrates that this method enables users to score better than baseline methods on one of two agent validation tasks.