Timezone: »

 
Poster
TokenLearner: Adaptive Space-Time Tokenization for Videos
Michael Ryoo · AJ Piergiovanni · Anurag Arnab · Mostafa Dehghani · Anelia Angelova

Thu Dec 09 08:30 AM -- 10:00 AM (PST) @ Virtual
in Poster Session 6 »

In this paper, we introduce a novel visual representation learning which relies on a handful of adaptively learned tokens, and which is applicable to both image and video understanding tasks. Instead of relying on hand-designed splitting strategies to obtain visual tokens and processing a large number of densely sampled patches for attention, our approach learns to mine important tokens in visual data. This results in efficiently and effectively finding a few important visual tokens and enables modeling of pairwise attention between such tokens, over a longer temporal horizon for videos, or the spatial content in image frames. Our experiments demonstrate strong performance on several challenging benchmarks for video recognition tasks. Importantly, due to our tokens being adaptive, we accomplish competitive results at significantly reduced computational cost. We establish new state-of-the-arts on multiple video datasets, including Kinetics-400, Kinetics-600, Charades, and AViD.

Keywords: Transformers · Representation Learning

Author Information

Michael Ryoo (Google; Stony Brook University)
AJ Piergiovanni (Indiana University)
Anurag Arnab (University of Oxford)
Mostafa Dehghani (Google Brain)
Anelia Angelova (Google Research)

More from the Same Authors