CompAct : COMPression of mAssively produCed visual daTa

In a context of massively produced and viewed visual data, the main objective of the COMPACT team is to develop data compression and processing algorithms based on strong theoretical foundations. Leveraging its expertise in signal and image processing, information theory, and statistical machine learning, the team addresses the problem of compression comprehensively. Thus, compression and processing algorithms are proposed, accompanied by theoretical guarantees regarding content preservation and information theoretical compression bounds.

Scientific challenges

1. Reducing the size of each individual visual data,
2. Reducing the size of a collection of visual data,
3. Reducing energy consumption.

The main research axes are

• Axis 1. Compressing for specific types of visual data, receivers and media,
• Axis 2. Sobriety for visual data,
• Axis 3. Acquisition/representation/processing co-design,
• Axis 4. Learning methods and guarantees.