Enabling Next-Generation Consumer Experience with Feature Coding for Machines

TL;DR

This paper introduces the Feature Coding for Machines (FCM) standard, enabling efficient neural network feature transmission for AI applications on consumer devices, significantly reducing data bitrate without sacrificing accuracy.

Contribution

It presents the FCM standard as part of MPEG-AI, facilitating efficient feature extraction, compression, and transmission for AI-driven applications on low-powered devices.

Findings

Reduces bitrate by 75.90% compared to remote inference

Maintains accuracy while improving data transfer efficiency

Supports AI applications on resource-constrained devices

Abstract

As consumer devices become increasingly intelligent and interconnected, efficient data transfer solutions for machine tasks have become essential. This paper presents an overview of the latest Feature Coding for Machines (FCM) standard, part of MPEG-AI and developed by the Moving Picture Experts Group (MPEG). FCM supports AI-driven applications by enabling the efficient extraction, compression, and transmission of intermediate neural network features. By offloading computationally intensive operations to base servers with high computing resources, FCM allows low-powered devices to leverage large deep learning models. Experimental results indicate that the FCM standard maintains the same level of accuracy while reducing bitrate requirements by 75.90% compared to remote inference.