Machine learning technique – using a SVM (Support Vector Machine) multi-classification scheme – ais used to improve the R-D model prediction accuracy for inter-frame CTUs by exploiting the potential information for the R-D relationships hidden in the video coding data.

The invention relates to a video quality determination method, and a related system implementing the method.

This patent presents an external memory-augmented network for low-light image enhancement.

This patent presents a key point-based salient instance segmentation network (KGDC-Net).

This patent presents a system that models task features by decoupling task-invariant and task-variant components using multi-level contrastive learning.

This patent presents a Local and Global Frequency feature-based Model (LGFM) for full-reference HDR image quality assessment.