Sorry, you need to enable JavaScript to visit this website.

facebooktwittermailshare

CASCADED TEMPORAL SPATIAL FEATURES FOR VIDEO ACTION RECOGNITION

Abstract: 

Extracting spatial-temporal descriptors is a challenging task for video-based human action recognition. We decouple the 3D volume of video frames directly into a cascaded temporal spatial domain via a new convolutional architecture. The motivation behind this design is to achieve deep nonlinear feature representations with reduced network parameters. First, a 1D temporal network with shared parameters is first constructed to map the video sequences along the time axis into feature maps in temporal domain. These feature maps are then organized into channels like those of RGB image (named as Motion Image here for abbreviation), which is desired to preserve both temporal and spatial information. Second, the Motion Image
is regarded as the input of the latter cascaded 2D spatial network. With the combination of the 1D temporal network and the 2D spatial network together, the size of whole network parameters is largely reduced. Benefiting from the Motion Image, our network is an end-to-end system for the task of action recognition, which can be trained with the classical algorithm of back propagation. Quantities of comparative experiments on two benchmark datasets demonstrate the effectiveness of our new architecture.

up
0 users have voted:

Paper Details

Authors:
Tingzhao Yu, Huxiang Gu, Lingfeng Wang, Shiming Xiang, Chunhong Pan
Submitted On:
15 September 2017 - 5:00am
Short Link:
Type:
Presentation Slides
Event:
Presenter's Name:
Tingzhao Yu
Paper Code:
1338
Document Year:
2017
Cite

Document Files

CTS_Tsingzao_v1_pdf.pdf

(36 downloads)

Subscribe

[1] Tingzhao Yu, Huxiang Gu, Lingfeng Wang, Shiming Xiang, Chunhong Pan, "CASCADED TEMPORAL SPATIAL FEATURES FOR VIDEO ACTION RECOGNITION", IEEE SigPort, 2017. [Online]. Available: http://sigport.org/2111. Accessed: Dec. 18, 2017.
@article{2111-17,
url = {http://sigport.org/2111},
author = {Tingzhao Yu; Huxiang Gu; Lingfeng Wang; Shiming Xiang; Chunhong Pan },
publisher = {IEEE SigPort},
title = {CASCADED TEMPORAL SPATIAL FEATURES FOR VIDEO ACTION RECOGNITION},
year = {2017} }
TY - EJOUR
T1 - CASCADED TEMPORAL SPATIAL FEATURES FOR VIDEO ACTION RECOGNITION
AU - Tingzhao Yu; Huxiang Gu; Lingfeng Wang; Shiming Xiang; Chunhong Pan
PY - 2017
PB - IEEE SigPort
UR - http://sigport.org/2111
ER -
Tingzhao Yu, Huxiang Gu, Lingfeng Wang, Shiming Xiang, Chunhong Pan. (2017). CASCADED TEMPORAL SPATIAL FEATURES FOR VIDEO ACTION RECOGNITION. IEEE SigPort. http://sigport.org/2111
Tingzhao Yu, Huxiang Gu, Lingfeng Wang, Shiming Xiang, Chunhong Pan, 2017. CASCADED TEMPORAL SPATIAL FEATURES FOR VIDEO ACTION RECOGNITION. Available at: http://sigport.org/2111.
Tingzhao Yu, Huxiang Gu, Lingfeng Wang, Shiming Xiang, Chunhong Pan. (2017). "CASCADED TEMPORAL SPATIAL FEATURES FOR VIDEO ACTION RECOGNITION." Web.
1. Tingzhao Yu, Huxiang Gu, Lingfeng Wang, Shiming Xiang, Chunhong Pan. CASCADED TEMPORAL SPATIAL FEATURES FOR VIDEO ACTION RECOGNITION [Internet]. IEEE SigPort; 2017. Available from : http://sigport.org/2111