Back to Search Document details
35th Meeting: Sapporo, JP, July 2024 2024-11-03 17:39
AhG12: Neural network-based intra prediction with DIMD mode derivation
Authors: Samuel Eadie Qualcomm Patrick Garus Qualcomm Thomas Ryder Qualcomm Pavel Nikitin Qualcomm Muhammed Coban Qualcomm Vadim Seregin Qualcomm Marta Karczewicz Qualcomm F. Urban InterDigital T. Dumas InterDigital Franck Galpin InterDigital Edouard François InterDigital
Abstract
In this contribution, NN-based intra prediction with the reduced model complexity is proposed. The models require up to 51.3% less MACs/pixels than the model in NNVC while preserving the coding gains. The following results are reported:
ARCHIVE :
JVET-AI0225-v9.docx JVET-AI0225-v9.pptx JVET-AI0225-v9.xlsm
Citation