Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

LLMs are Good Action Recognizers

About

Skeleton-based action recognition has attracted lots of research attention. Recently, to build an accurate skeleton-based action recognizer, a variety of works have been proposed. Among them, some works use large model architectures as backbones of their recognizers to boost the skeleton data representation capability, while some other works pre-train their recognizers on external data to enrich the knowledge. In this work, we observe that large language models which have been extensively used in various natural language processing tasks generally hold both large model architectures and rich implicit knowledge. Motivated by this, we propose a novel LLM-AR framework, in which we investigate treating the Large Language Model as an Action Recognizer. In our framework, we propose a linguistic projection process to project each input action signal (i.e., each skeleton sequence) into its ``sentence format'' (i.e., an ``action sentence''). Moreover, we also incorporate our framework with several designs to further facilitate this linguistic projection process. Extensive experiments demonstrate the efficacy of our proposed framework.

Haoxuan Qu, Yujun Cai, Jun Liu• 2024

Related benchmarks

TaskDatasetResultRank
Action RecognitionNTU RGB+D 120 (X-set)
Accuracy91.5
770
Action RecognitionNTU RGB+D (Cross-View)
Accuracy98.4
652
Action RecognitionNTU RGB+D 60 (X-sub)
Accuracy95
496
Action RecognitionNTU RGB+D X-sub 120
Accuracy88.7
473
Action RecognitionNTU RGB+D X-View 60
Accuracy98.4
218
Action RecognitionToyota SmartHome (TSH) (CV1)
Accuracy36.1
68
Action RecognitionNTU RGB+D Xsub 60 (Cross-Subject 55/5)
Accuracy95
66
Action RecognitionNTU-RGB+D (X-Sub)--
62
Action RecognitionNTU-RGBD 120 (xsub)
Accuracy88.7
24
Human Action RecognitionUAV-Human X-Sub
Accuracy46.3
15
Showing 10 of 12 rows

Other info

Code

Follow for update