Compare UpSampling2D + Conv2D with Conv2DTranspose

[pesekon2]

Compare results reached with Conv2DTranspose used for upsampling with results reached with UpSampling2D and Conv2D used for upsampling.

Before the test, Conv2DTranspose will be used as it is generally considered better.

Links touching the topic:

• https://distill.pub/2016/deconv-checkerboard/
• https://towardsdatascience.com/types-of-convolutions-in-deep-learning-717013397f4d
• https://www.machinecurve.com/index.php/2019/12/11/upsampling2d-how-to-use-upsampling-with-keras/