Video Coding for Machines: Partial Transmission of SIFT Features

This research focuses on video coding for machines, specifically utilizing local feature descriptors known as SIFT (Scale-Invariant Feature Transform). The primary goal is to enable support for machine learning algorithms through video encoding. The study investigates partial transmission of SIFT features, addressing the challenge of efficient video encoding and transmission under bandwidth limitations while maintaining usability for machine learning applications. Additionally, it explores the use of hierarchical coding methods to optimize transmission efficiency and investigates leveraging SIFT features for more efficient video processing within end-to-end machine learning systems.