- Dong, C., Loy, C. C., He, K., & Tang, X. (2014, September). Learning a deep convolutional network for image super-resolution. In European conference on computer vision (pp. 184-199). Springer, Cham.
- Low to high res
- Path extraction+aggregation -> conv
- Non linear mapping -> Extract n1 dimensional features, map each to n2 dimensional. (1x1)
- Reconstruction -> predicted overlapping are averaged
- MSE loss