From smarter images to trusted heritage: JPEG in the age of AI

Digital images are no longer passive documents: they are increasingly created, modified, compressed, enhanced, analysed and transformed using artificial intelligence. For archives and heritage institutions, this evolution creates significant opportunities, while also raising questions concerning their authenticity, intellectual property rights, authorship, ownership, trustworthiness and intrinsic value.
This keynote will examine the two complementary ways in which JPEG is responding to the emergence of AI.
First, we will introduce JPEG AI, the first international standard for end-to-end learning-based image coding, which uses AI to achieve efficient compression and create a compact representation suitable for both human viewing and machine processing. As an example, its compressed-domain capabilities can support operations such as on-the-fly image enhancement, analysis and computer vision without requiring the complete image to be first reconstructed or the archived version to be altered or duplicated.
Second, the keynote will present JPEG Trust, an international standard addressing the growing need for transparency regarding how an image has been created, modified, attributed and deployed.
JPEG Trust provides a versatile framework for recording and assessing provenance, authenticity, integrity, authorship and intellectual property information throughout the media life cycle.
Together, JPEG AI and JPEG Trust illustrate how recent JPEG formats can combine advanced functionality with the transparency and accountability required by a wide range of applications, including archival and cultural heritage.