Action Prediction Using Unsupervised Semantic Reasoning

Liu, Cuiwei; Lu, Yaguang; Shi, Xiaohu; Li, Zhaokui; Zhao, Liang

doi:10.1007/978-3-319-70090-8_50

Cited by 1 publication

(1 citation statement)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• A latent discriminative action prediction model that indicates the progress state of an incomplete video with a latent variable and accomplishes precise recognition of actions by jointly learning the relationship among actions, observed concepts, inferred concepts, latent progress states, and video observations. This paper is an extension of our previous work [19]. These extensions include: a refined double-layer learning framework for action prediction; a new discriminative action prediction model with latent variables; a new scheme of video representations; and more experimental results on three datasets.…”

Section: The Main Contributions Of This Paper Includementioning

confidence: 97%

A Novel Double-Layer Framework for Action Prediction

Liu

Shi

et al. 2019

IEEE Access

Self Cite

View full text Add to dashboard Cite

Action prediction aims to infer the category of an action before it is fully executed. It is a challenging task since neither sufficient discriminative information nor the definite progress state of action can be obtained in an incomplete video. In this paper, we propose a novel double-layer learning framework for predicting the category of action from partial observations. Particularly, in the first layer of the framework, an unsupervised semantic reasoning method is presented for exploiting semantic information of an input incomplete video as well as inferring the future semantic information using the prior knowledge provided by training full videos. In the second layer of the framework, a discriminative action prediction model introduces a latent variable to indicate the progress state of the input video and captures the relationship among the actions, video observations, the semantic information, and the latent progress state for predicting the action label of the input video. Extensive experimental results on UT-I #1, UT-I #2, and UCF Sports datasets demonstrate the superiority of our method in predicting actions at the early stage.

show abstract

Section: The Main Contributions Of This Paper Includementioning

confidence: 97%