Is Image Super-Resolution Helpful for Other Vision Tasks?

Is Image Super-resolution Helpful for Other Vision Tasks?

Dengxin Dai, Yujian Wang, Yuhua Chen, and Luc Van Gool

Summary

Six state-of-the-art image super-resolution (ISR) methods are evaluated on four popular vision tasks, namely edge detection, semantic image segmentation, digit recognition and scene recognition.
ISR is helpful for other vision tasks when the input images are of low-resolution.
The standard perceptual criteria, such as PSNR and SSIM, correlate well with the usefulness of ISR to these vision tasks.

Code for scene classification is provided. Code for other tasks will be available soon.

code for scene classification

Paper

slide

poster

BibTex

Dengxin Dai, Yujian Wang, Yuhua Chen, and Luc Van Gool. Is Image Super-resolution Helpful for Other Vision Tasks? In WACV 2016.