Awesome

Bringing Alive Blurred Moments

Oral presentation at IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019. Link to paper

Designed a deep convolutional architecture to extract a sharp video from a motion blurred image. The first stage involves unsupervised training of a novel spatiotemporal network for motion extraction from short video sequences. The above network is utilized for guided training of a CNN which extracts the same motion embedding from a single blurred image. The above networks are finally linked with our efficient deblurring network to generate the sharp video. Our framework delivers state-of-the-art accuracy in single image deblurring and video extraction while being faster and more compact. The work is among the first of its kind as the traditional algorithms extract only a single deblurred image.

Link to deblurring Results on GoPro Benchmark's test images: Google Drive

DEMO: Blurred Images: alt text

Video Extraction Results: Left : Generated Video Right : Ground Truth Video alt text

Citation

If you find our paper/results helpful in your research or work please cite our paper.

@inproceedings{purohit2019bringing,
  title={Bringing alive blurred moments},
  author={Purohit, Kuldeep and Shah, Anshul and Rajagopalan, AN},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={6830--6839},
  year={2019}
}