近年来,面向机器视觉视频的研究和应用越来越广泛,这对此类视频的存储和传输都提出了巨大的挑战。视频编码标准如多功能视频编码(Versatile Video Coding,VVC)能实现高效的全分辨率压缩与重建,但是对机器视觉任务而言,这种压缩方法是有...近年来,面向机器视觉视频的研究和应用越来越广泛,这对此类视频的存储和传输都提出了巨大的挑战。视频编码标准如多功能视频编码(Versatile Video Coding,VVC)能实现高效的全分辨率压缩与重建,但是对机器视觉任务而言,这种压缩方法是有冗余的。因此,提出了一种在VVC编码过程中结合显著性检测的视频编码方法用于机器任务,用实例分割网络掩膜基于区域的卷积神经网络(Mask Region-based Convolutional Neural Network,Mask R-CNN)获得包含对象的二进制掩膜,并依此判定是否为感兴趣区域,指导VVC编码过程中编码树单元(Coding Tree Unit,CTU)的量化参数的偏移。实验证明,与VVC基线方法相比,所提方法可以在相似的检测精度下节省一定的比特率。展开更多
A complementarity hypothesis concerning outsider and insider perspectives of a gargantuan black hole is proposed. The two thought experiments presented herein are followed by a brief discussion of a new interpretation...A complementarity hypothesis concerning outsider and insider perspectives of a gargantuan black hole is proposed. The two thought experiments presented herein are followed by a brief discussion of a new interpretation of black hole interior “space-and-time-reversal”. Specifically, it is proposed that the “singularity” space of the black hole interior is time-like and the expansion time of the black hole interior is space-like. The resemblance of this new insider interpretation to our own expanding and redshifting big bang universe is compelling.展开更多
文摘近年来,面向机器视觉视频的研究和应用越来越广泛,这对此类视频的存储和传输都提出了巨大的挑战。视频编码标准如多功能视频编码(Versatile Video Coding,VVC)能实现高效的全分辨率压缩与重建,但是对机器视觉任务而言,这种压缩方法是有冗余的。因此,提出了一种在VVC编码过程中结合显著性检测的视频编码方法用于机器任务,用实例分割网络掩膜基于区域的卷积神经网络(Mask Region-based Convolutional Neural Network,Mask R-CNN)获得包含对象的二进制掩膜,并依此判定是否为感兴趣区域,指导VVC编码过程中编码树单元(Coding Tree Unit,CTU)的量化参数的偏移。实验证明,与VVC基线方法相比,所提方法可以在相似的检测精度下节省一定的比特率。
文摘A complementarity hypothesis concerning outsider and insider perspectives of a gargantuan black hole is proposed. The two thought experiments presented herein are followed by a brief discussion of a new interpretation of black hole interior “space-and-time-reversal”. Specifically, it is proposed that the “singularity” space of the black hole interior is time-like and the expansion time of the black hole interior is space-like. The resemblance of this new insider interpretation to our own expanding and redshifting big bang universe is compelling.
