Title
Impact of temporal subsampling on accuracy and performance in practical video classification.
Abstract
In this paper we evaluate three state-of-the-art neural-network-based approaches for large-scale video classification, where the computational efficiency of the inference step is of particular importance due to the ever increasing amount of data throughput for video streams. Our evaluation focuses on finding good efficiency vs. accuracy tradeoffs by evaluating different network configurations and parameterizations. In particular, we investigate the use of different temporal subsampling strategies, and show that they can be used to effectively trade computational workload against classification accuracy. Using a subset of the YouTube-8M dataset, we demonstrate that workload reductions in the order of 10 X, 50 X and 100 X can be achieved with accuracy reductions of only 1.3 %, 6.2 % and 10.8 %, respectively. Our results show that temporal subsampling is a simple and generic approach that behaves consistently over the considered classification pipelines and which does not require retraining of the underlying networks.
Year
Venue
Field
2017
European Signal Processing Conference
Data mining,Pipeline transport,Workload,Computer science,Inference,Feature extraction,Artificial intelligence,Throughput,Artificial neural network,Machine learning,Retraining
DocType
ISSN
Citations 
Conference
2076-1465
2
PageRank 
References 
Authors
0.36
16
6
Name
Order
Citations
PageRank
Florian Scheidegger1154.01
Cavigelli, L.224422.75
Michael Schaffner36812.41
A Cristiano I Malossi4659.29
Constantine Bekas5496.59
Luca Benini6131161188.49