Image Coding

Image coding aims to efficiently represent images in digital form, optimizing for either human perception or machine analysis (Image Coding for Machines, or ICM). Current research emphasizes developing versatile codecs that adapt to various bitrates, downstream tasks, and quality requirements, often employing neural networks with architectures like autoencoders, transformers, and vector quantization. These advancements are crucial for improving the efficiency of image transmission and storage, particularly in applications like AI-driven image analysis and 3D modeling, where minimizing bandwidth and computational costs is paramount.