MadEye: Boosting Live Video Analytics Accuracy with Adaptive Camera Configurations

04/04/2023
by   Mike Wong, et al.
0

Camera orientations (i.e., rotation and zoom) govern the content that a camera captures in a given scene, which in turn heavily influences the accuracy of live video analytics pipelines. However, existing analytics approaches leave this crucial adaptation knob untouched, instead opting to only alter the way that captured images from fixed orientations are encoded, streamed, and analyzed. We present MadEye, a camera-server system that automatically and continually adapts orientations to maximize accuracy for the workload and resource constraints at hand. To realize this using commodity pan-tilt-zoom (PTZ) cameras, MadEye embeds (1) a search algorithm that rapidly explores the massive space of orientations to identify a fruitful subset at each time, and (2) a novel knowledge distillation strategy to efficiently (with only camera resources) select the ones that maximize workload accuracy. Experiments on diverse workloads show that MadEye boosts accuracy by 2.9-25.7 resource usage, or achieves the same accuracy with 2-3.7x lower resource costs.

READ FULL TEXT

page 5

page 6

page 10

page 11

research
11/03/2018

ReXCam: Resource-Efficient, Cross-Camera Video Analytics at Enterprise Scale

The deployment of large camera networks for video analytics is an establ...
research
01/19/2022

GEMEL: Model Merging for Memory-Efficient, Real-Time Video Analytics at the Edge

Video analytics pipelines have steadily shifted to edge deployments to r...
research
04/25/2021

Swimmer Stroke Rate Estimation From Overhead Race Video

In this work, we propose a swimming analytics system for automatically d...
research
08/19/2023

ILCAS: Imitation Learning-Based Configuration-Adaptive Streaming for Live Video Analytics with Cross-Camera Collaboration

The high-accuracy and resource-intensive deep neural networks (DNNs) hav...
research
02/03/2021

Llama: A Heterogeneous Serverless Framework for Auto-Tuning Video Analytics Pipelines

The proliferation of camera-enabled devices and large video repositories...
research
06/29/2022

Turbo: Opportunistic Enhancement for Edge Video Analytics

Edge computing is being widely used for video analytics. To alleviate th...
research
05/11/2021

DeepLight: Robust Unobtrusive Real-time Screen-Camera Communication for Real-World Displays

The paper introduces a novel, holistic approach for robust Screen-Camera...

Please sign up or login with your details

Forgot password? Click here to reset