Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit

ECS795P Deep Learning and Computer Vision, 2023

Course Work 1: Image Super-resolution Using Deep Learning

1. Suppose the settings of a SRCNN as: f1=9, f2=3, f3=5, how many pixels of the low-resolution image are utilized to reconstruct a pixel of the high-resolution image with the SRCNN?  (10% of CW1)

2. Why the deep convolutional neural network is superior to perform image super-resolution? Give one reason to explain it. (10% of CW1)

3. 1) In the context of image super-resolution, explain the definition (how to calculate) of peak signal-to-noise ratio (PSNR) and why can PSNR be applied to measure the quality of output images?

    2) Show the ground truth (GT) image, and the high-resolution images by interpolation (HR-Base) and SRCNN (HR-SRCNN). Also put the PSNR values below the high-resolution images. (10% of CW1)