The PCS 2024 technical program will include a series of Special Sessions that complement the regular sessions by focusing on important emerging research topics. Interested authors may submit papers to these special sessions via the submission site.


Advances in Next Generation Video Coding

The advances of next generation video coding have two trends. One is to use deep learning or reinforcement learning to further optimize coding tools of traditional video coding, such as intra/inter prediction, loop filtering, and encoder optimization. The other is the end-to-end video coding framework, which uses neural networks to fully replace the traditional video coding. In this way, the neural networks of end-to-end video coding framework could be jointly optimized. In this special session, we aim at gathering together and discussing some of the most recent and significant results on advances in next generation video coding, which we believe is very important for the CAS society. Research papers are solicited in, but not limited to the following topics:

  • Deep learning-based coding tools for traditional video coding
  • Reinforcement leaning-based encoder optimization
  • End-to-end video coding frameworks for low delay scenarios
  • Novel end-to-end video coding frameworks for random access scenarios





Semantic Visual Compression towards Machine and Human Vision

Motivated by the recent advances of artificial intelligence (AI) for visual processing and understanding, semantic visual compression towards machine and human vision is becoming increasingly important. Compared with signal-level rate-distortion oriented visual compression, semantic visual compression is optimized towards semantic-level rate-distortion, preserving the fidelity in terms of machine analysis and human perception, which is highly desired in various AI-based applications, such as intelligent traffic, smart city, and Industry 4.0 systems. However, many challenges remain in developing effective and practical semantic visual compression targeted towards machine and human vision. This special session aims to present state-of-the-art methodologies and techniques, fostering advancements in the semantic visual compression for both machine and human vision.

This special session solicits the contributions for semantic compression towards machine and human vision. Topics of interest include:

  • Semantic visual compression towards single/multiple machine vision task(s)
  • Semantic visual compression towards machine and/or human vision
  • Joint semantic visual compression of multi-modal data, e.g., natural, infrared, point cloud and under water data, etc.
  • Semantic compression quality assessment for machine and/or human vision
  • Privacy-protected visual compression
  • Novel semantic compression frameworks and datasets





Realistic 3D Graphics Representations and Compression

In an era where the digital world is ever-expanding, the need for realistic 3D graphics has permeated every corner of technology, from entertainment and education to scientific visualization and manufacturing. The burgeoning demand for high-fidelity, detailed 3D content comes with significant challenges in storage, bandwidth, and processing power. This special session aims to address these challenges by exploring advanced 3D graphics representations and compression techniques that enable efficient creation, storage, and transmission without compromising on quality. Compression technologies for 3D graphics are not merely a matter of storage economy; they are essential for the accessibility and usability of 3D data across different platforms and devices. They enable real-time rendering for interactive applications, facilitate the streaming of complex scenes for remote collaboration, and are critical for the scalability of 3D content in networked environments. Hence, this session seeks contributions that push the envelope in compression algorithms, offering insights that could potentially revolutionize how we handle 3D data.

We invite original, high-quality submissions on topics including but not limited to:

  • Advances in 3D mesh compression
  • Novel point cloud compression algorithms
  • Efficiency and quality metrics for 3D representation
  • Real-time compression techniques for interactive graphics
  • Machine learning approaches to 3D graphics compression
  • Standards for 3D graphics (e.g., MPEG-4, MPEG-I, glTF)
  • Compression techniques for 3D printing data
  • Data structures and algorithms for large-scale 3D data
  • Transmission and streaming of 3D content
  • 3D graphics in augmented and virtual reality
  • Visualization and manipulation of compressed 3D data
  • Security and intellectual property issues in the compression of 3D graphics
  • Case studies and industry-specific applications (e.g., gaming, film, virtual tours)