Low-complexity intra coding algorithm based on convolutional neural network for HEVC

In this paper, we propose a fast coding unit (CU) size decision algorithm for high efficiency video coding (HEVC) based on convolutional neural network. The proposed fast algorithm contributes to decrease no less than two CU partition modes in each coding tree unit for full rate-distortion optimization processing, thereby reducing the encoder hardware complexity. Moreover, our algorithm use only texture information and it does not depend on the correlations among CU depths or spatially nearby CUs. It is friendly to the parallel processing and it can improve the pipeline process of RDO. The proposed algorithm is implemented in the reference software of HEVC (HM16.7). The simulation results show that the proposed algorithm can achieve over 67.3% computation complexity reduction comparing to the original HEVC algorithm.