Single Image Super-resolution via Dense Blended Attention Generative Adversarial Network for Clinical Diagnosis
This is an incremental improvement for clinical diagnosis, but it has been abandoned in favor of more general work.
The authors addressed the problem of high GPU memory usage and low GPU utilization in super-resolution training by abandoning DenseNet, but the work was limited to medical images and has been superseded by a newer general image application.
During training phase, more connections (e.g. channel concatenation in last layer of DenseNet) means more occupied GPU memory and lower GPU utilization, requiring more training time. The increase of training time is also not conducive to launch application of SR algorithms. This's why we abandoned DenseNet as basic network. Futhermore, we abandoned this paper due to its limitation only applied on medical images. Please view our lastest work applied on general images at arXiv:1911.03464.