Recent News

  • [2022-01-01] Note that the Stereo Blur dataset is publicly available all the time.
  • [2022-01-01] Original stereo videos are no longer publicly available. If you want to download them or build a new dataset upon the original sequences, you need to fill in this form and then apply for access with us via email.
  • [2019-11-08] Original stereo videos sequences are publicly available now.
  • [2019-04-01] Stereo Blur Dataset initial release.

The datasets on this page can not be used for commercial purposes. If your research use, transform, or build upon our work, please consider citing.
Stereo Blur Dataset / Blurry


The Stereo Blur Dataset is a large-scale multi-scene dataset for stereo deblurring in dynamic scenes (both indoor and outdoor). It contains 20,637 blurry-sharp stereo images from 135 diverse video clips (480 fps).

  • Training: 17,319 stereo images (98 videos)
  • Testing: 3,318 stereo images (37 videos)
  • Key Features: Large-scale, Multi-scene, Stereo Blur
  • Shot Fashions: Handheld, Fixed, Onboard
  • Camera: ZED Camera (720p, 60fps)
  • Location: Shenzhen, China

Explore the Dataset

Get Started with Stereo Blur Dataset!


DAVANet: Stereo Deblurring with View Aggregation
  • Shangchen Zhou
  • Jiawei Zhang
  • Wangmeng Zuo
  • Haozhe Xie
  • Jinshan Pan
  • Jimmy Ren

The 2019 Conference on Computer Vision and Pattern Recognition (CVPR 2019, Oral)

Team Members