Back to Search Document details
41st Meeting: by teleconference, CH, January 2026 2026-01-14 18:58
Crosscheck of JVET-AO0108 (EE1-4.1: Deep Reference Frame Generation for Inter Prediction Enhancement with Structural Re-parameterization)
Authors: Nisha Bhaskar Huawei Elena Alshina
Abstract
This contribution reports the inference and training crosschecks of JVET-AO0108, EE1-4.1[1]. JVET-AO0108 studies the effect of using different loss functions to train the large and small Deep Generation Frame Generation (DRFG) networks which were presented in JVET-AN0195[2] and JVET-AN0193[3], respectively.
Meeting Notes
JVET-AO0274 Crosscheck of JVET-AO0108 (EE1-4.1: Deep Reference Frame Generation for Inter Prediction Enhancement with Structural Re-parameterization) N. Bhaskar (Huawei) [late]
References:
JVET-AO0108
ARCHIVE :
EE1-4.1.1-DRF-HOP_vs_NNVC-15-Cx.xlsm EE1-4.1.1-DRF-HOP_vs_NNVC-15-TrainingCx.xlsm EE1-4.1.2-DRF-LOP_vs_NNVC-15-Cx.xlsm EE1-4.1.2-DRF-LOP_vs_NNVC-15-TrainingCx.xlsm EE1-4.1.3-DRF-LOP_vs_NNVC-15-Cx.xlsm EE1-4.1.3-DRF-LOP_vs_NNVC-15-TrainingCx.xlsm JVET-AO0274-v3-clean.docx JVET-AO0274-v3.docx ~$EE1-4.1.2-DRF-LOP_vs_NNVC-15-TrainingCx.xlsm
Decisions
JVET-AO0274 Crosscheck of JVET-AO0108 (EE1-4.1: Deep Reference Frame Generation for Inter Prediction Enhancement with Structural Re-parameterization) N. Bhaskar (Huawei) [late]
Citation