Fig. 4: Experiment results with BCNN in SPI with the MNIST dataset at 16× and 64× compression ratios.
From: Approximating the uncertainty of deep learning reconstruction predictions in single-pixel imaging
a Ten representative ground-truth images from the MNIST testing dataset. b Input images and predictions of BCNN in SPI at 16× compression ratio. c Input images and predictions of BCNN in SPI at 64× compression ratio.
