Back to Search Document details
42nd Meeting: Santa Eulària, ES, April 2026 2026-04-24 15:40
EE1-4: DCVC-RT as a Learned End-to-End Intra Frame Codec
Authors: Mouad Aderdor Huawei Timofey Solovyev Huawei Elena Alshina Huawei F. Urban InterDigital Franck Galpin InterDigital Edouard François InterDigital
Abstract
This contribution reports EE1-4 tests. In total, 4 experiments were conducted where DCVC-RT is used as an external coder, which feeds I-frames to VTM using the multiple hybrid frameworks and methods. Two frameworks were tested, single layer and multilayer. Two methods of insertion were tested on both frameworks; always inserting DCVC-RT and insertion based on frame level RDO. DCVC-RT’s decoder has a complexity of 365 kMACs/pixel and 36.4 M Parameters.
Meeting Notes
JVET-AP0080 EE1-4: DCVC-RT as a Learned End-to-End Intra Frame Codec [M. Aderdor, T. Solovyev, E. Alshina (Huawei), F. Urban, F. Galpin, E. François (InterDigital)]
ARCHIVE :
EE1-4.1_VTM_vs_DCVC-RT.xlsm EE1-4.2_VTM_vs_DCVC-RT-FLS.xlsm EE1-4.3_VTM_vs_DCVC-RT-Single-LMCS0.xlsm EE1-4.4_VTM_vs_DCVC-RT-FLS-Single-LMCS0.xlsm JVET-AP00080_v4.docx
Decisions
JVET-AP0080 EE1-4: DCVC-RT as a Learned End-to-End Intra Frame Codec [M. Aderdor, T. Solovyev, E. Alshina (Huawei), F. Urban, F. Galpin, E. François (InterDigital)]
Citation