Free Porn
Thursday, July 18, 2024

MPEG Issues Call for Evidence for Video Coding for Machines

At the recent 139th MPEG meeting, MPEG Technical Requirements (WG 2) issued a Call for Evidence (CfE) for technologies and solutions enabling efficient feature coding for machine vision tasks.

MPEG’s exploration work on Video Coding for Machines aims at compressing features for machine-performed tasks such as video object detection and event analysis. As neural networks increase in complexity, architectures such as collaborative intelligence, whereby a network is distributed across an edge device and the cloud, become advantageous. With the rise of newer network architectures being deployed amongst a heterogenous population of edge devices, such architectures bring flexibility to systems implementers.

Due to such architectures, there is a need to efficiently compress intermediate feature information for transport over wide area networks (WANs). As feature information differs substantially from conventional image or video data, coding technologies and solutions for machine usage could differ from conventional human-viewing-oriented applications to achieve optimised performance. With the rise of machine learning technologies and machine vision applications, the amount of video and images consumed by machines has rapidly grown.

Typical use cases include intelligent transportation, smart city technology, intelligent content management, etc., which incorporate machine vision tasks such as object detection, instance segmentation, and object tracking. Due to the large volume of video data, extracting and compressing the feature from a video is essential for efficient transmission and storage. Feature compression technology solicited in this CfE can also be helpful in other regards, such as computational offloading and privacy protection.

Over the last three years, MPEG has investigated potential technologies for efficiently compressing feature data for machine vision tasks and established an evaluation mechanism that includes feature anchors, rate-distortion-based metrics, and evaluation pipelines.

Discussion of the submissions in response to the CfE will take place at the 140th MPEG meeting in October 2022.


Related Articles

Stay Connected


Latest Articles