Rate-Accuracy Trade-Off In Video Classification With Deep Convolutional Neural Networks

TL;DR

This paper explores how to reduce bandwidth in video classification by using compressed video data directly, achieving significant bitrate savings with minimal accuracy loss across different CNN models and datasets.

Contribution

It introduces a method to perform CNN-based video classification directly on compressed video data, reducing bandwidth without complex pre-processing.

Findings

Achieved 11%-94% bitrate reduction with minimal accuracy impact.

Using motion vectors and texture info at low bitrates enables effective classification.

Potential for classification at as low as 3 kbps with acceptable accuracy loss.

Abstract

Advanced video classification systems decode video frames to derive the necessary texture and motion representations for ingestion and analysis by spatio-temporal deep convolutional neural networks (CNNs). However, when considering visual Internet-of-Things applications, surveillance systems and semantic crawlers of large video repositories, the video capture and the CNN-based semantic analysis parts do not tend to be co-located. This necessitates the transport of compressed video over networks and incurs significant overhead in bandwidth and energy consumption, thereby significantly undermining the deployment potential of such systems. In this paper, we investigate the trade-off between the encoding bitrate and the achievable accuracy of CNN-based video classification models that directly ingest AVC/H.264 and HEVC encoded videos. Instead of retaining entire compressed video bitstreams and applying complex optical flow calculations prior to CNN processing, we only retain motion vector and select texture information at significantly-reduced bitrates and apply no additional processing prior to CNN ingestion. Based on three CNN architectures and two action recognition datasets, we achieve 11%-94% saving in bitrate with marginal effect on classification accuracy. A model-based selection between multiple CNNs increases these savings further, to the point where, if up to 7% loss of accuracy can be tolerated, video classification can take place with as little as 3 kbps for the transport of the required compressed video information to the system implementing the CNN models.