An Investigation into Performance Factors of Two-Stream I3D Networks

Chuan Dai, Yajuan Wei, Zhijie Xu, Minsi Chen, Ying Liu, Jiulun Fan

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Two-Stream Inflated 3D ConvNet (I3D) is based on 2D convolutional networks. It is inflated into 3D to deal with spatiotemporal feature extraction and classification in videos. I3D network is an efficient solution for video action recognition, and outstanding results have been obtained after applying the model pre-trained with Kinetics dataset. This paper discusses some ideas for improving the efficiency of I3D network. Instead of counting on network architecture improvement, efforts are focused on two aspects: 1) from the point of view of data pre-processing, including training data cleansing and augmentation. A range of data augmentation schemes are investigated to enhance the balance and regularity of input data in the training and testing phases. This idea is inspired by the original I3D model and proposers. 2) from the perspective of network backbones, for example, through the application of ResNet-50 as an alternative backbone model to gain a better perception into key performance factors for Two-Stream I3D networks. Experiment results clearly show that the proposed hybrid improvement strategy brings substantial improvement in recognition accuracy for benchmark and practical datasets.

Original languageEnglish
Title of host publication2021 26th International Conference on Automation and Computing
Subtitle of host publicationSystem Intelligence through Automation and Computing, ICAC 2021
EditorsChenguang Yang
PublisherIEEE
Number of pages6
ISBN (Electronic)9781860435577
ISBN (Print)9781665443524
DOIs
Publication statusPublished - 15 Nov 2021
Event26th International Conference on Automation and Computing - University of Portsmouth, Portsmouth, United Kingdom
Duration: 2 Sep 20214 Sep 2021
Conference number: 26
http://www.cacsuk.co.uk/index.php/icac2021
https://www.ieee-ras.org/conferences-workshops/technically-co-sponsored/icac
https://ieeexplore.ieee.org/xpl/conhome/9594055/proceeding

Conference

Conference26th International Conference on Automation and Computing
Abbreviated titleICAC 2021
Country/TerritoryUnited Kingdom
CityPortsmouth
Period2/09/214/09/21
Internet address

Fingerprint

Dive into the research topics of 'An Investigation into Performance Factors of Two-Stream I3D Networks'. Together they form a unique fingerprint.

Cite this