End-to-End Learning for Video Frame Compression with Self-Attention

被引:3
|
作者
Zou, Nannan [2 ]
Zhang, Honglei [1 ]
Cricri, Francesco [1 ]
Tavakoli, Hamed R. [1 ]
Lainema, Jani [1 ]
Aksu, Emre [1 ]
Hannuksela, Miska [1 ]
Rahtu, Esa [2 ]
机构
[1] Nokia Technol, Espoo, Finland
[2] Tampere Univ, Tampere, Finland
关键词
D O I
10.1109/CVPRW50498.2020.00079
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
One of the core components of conventional (i.e., non-learned) video codecs consists of predicting a frame from a previously-decoded frame, by leveraging temporal correlations. In this paper, we propose an end-to-end learned system for compressing video frames. Instead of relying on pixel-space motion (as with optical flow), our system learns deep embeddings of frames and encodes their difference in latent space. At decoder-side, an attention mechanism is designed to attend to the latent space of frames to decide how different parts of the previous and current frame are combined to form the final predicted current frame. Spatially-varying channel allocation is achieved by using importance masks acting on the feature-channels. The model is trained to reduce the bitrate by minimizing a loss on importance maps and a loss on the probability output by a context model for arithmetic coding. In our experiments, we show that the proposed system achieves high compression rates and high objective visual quality as measured by MS-SSIM and PSNR. Furthermore, we provide ablation studies where we highlight the contribution of different components.
引用
收藏
页码:580 / 584
页数:5
相关论文
共 50 条
  • [1] End-to-End ASR with Adaptive Span Self-Attention
    Chang, Xuankai
    Subramanian, Aswin Shanmugam
    Guo, Pengcheng
    Watanabe, Shinji
    Fujita, Yuya
    Omachi, Motoi
    INTERSPEECH 2020, 2020, : 3595 - 3599
  • [2] END-TO-END NEURAL SPEAKER DIARIZATION WITH SELF-ATTENTION
    Fujita, Yusuke
    Kanda, Naoyuki
    Horiguchi, Shota
    Xue, Yawen
    Nagamatsu, Kenji
    Watanabe, Shinji
    2019 IEEE AUTOMATIC SPEECH RECOGNITION AND UNDERSTANDING WORKSHOP (ASRU 2019), 2019, : 296 - 303
  • [3] Self-Attention Transducers for End-to-End Speech Recognition
    Tian, Zhengkun
    Yi, Jiangyan
    Tao, Jianhua
    Bai, Ye
    Wen, Zhengqi
    INTERSPEECH 2019, 2019, : 4395 - 4399
  • [4] An End-to-End Learning Framework for Video Compression
    Lu, Guo
    Zhang, Xiaoyun
    Ouyang, Wanli
    Chen, Li
    Gao, Zhiyong
    Xu, Dong
    IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2021, 43 (10) : 3292 - 3308
  • [5] END-TO-END SPEECH SUMMARIZATION USING RESTRICTED SELF-ATTENTION
    Sharma, Roshan
    Palaskar, Shruti
    Black, Alan W.
    Metze, Florian
    2022 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2022, : 8072 - 8076
  • [6] Efficient decoding self-attention for end-to-end speech synthesis
    Zhao, Wei
    Xu, Li
    FRONTIERS OF INFORMATION TECHNOLOGY & ELECTRONIC ENGINEERING, 2022, 23 (07) : 1127 - 1138
  • [7] On the localness modeling for the self-attention based end-to-end speech synthesis
    Yang, Shan
    Lu, Heng
    Kang, Shiyin
    Xue, Liumeng
    Xiao, Jinba
    Su, Dan
    Xie, Lei
    Yu, Dong
    Neural Networks, 2020, 125 : 121 - 130
  • [8] Very Deep Self-Attention Networks for End-to-End Speech Recognition
    Ngoc-Quan Pham
    Thai-Son Nguyen
    Niehues, Jan
    Mueller, Markus
    Waibel, Alex
    INTERSPEECH 2019, 2019, : 66 - 70
  • [9] End-to-end Parking Behavior Recognition Based on Self-attention Mechanism
    Li, Penghua
    Zhu, Dechen
    Mou, Qiyun
    Tu, Yushan
    Wu, Jinfeng
    2023 2ND ASIA CONFERENCE ON ALGORITHMS, COMPUTING AND MACHINE LEARNING, CACML 2023, 2023, : 371 - 376
  • [10] On the localness modeling for the self-attention based end-to-end speech synthesis
    Yang, Shan
    Lu, Heng
    Kang, Shiyin
    Xue, Liumeng
    Xiao, Jinba
    Su, Dan
    Xie, Lei
    Yu, Dong
    NEURAL NETWORKS, 2020, 125 : 121 - 130