ContinuousSR
ContinuousSR copied to clipboard
peylnog
Official Code for “Pixel to Gaussian: Ultra-Fast Continuous Super-Resolution with 2D Gaussian Modeling”
Hello, and thanks for your innovative work! I truly appreciate you making the code and pretrained models publicly available, this greatly helps the community. While attempting to reproduce the results,...
Can u share your configs about training, base hat? 您好，感谢您的杰出工作。您可以分享您基于HAT的训练配置yml文件吗？我注意到公开的预训练权重中，HAT的upscale=4,您的意思是在任意倍数的训练中是固定了的上采样倍数吗？这里搞不太懂，恳切得到您的解答，谢谢
First of all, thank you for your excellent work and for making the code publicly available. I've learned a lot from your project. I am currently studying your paper and...
论文中协方差矩阵的定义以及统计中各个部分的取值范围 实际代码中我也看到cho1,cho2和cho3的定义是在这三个取值范围里均匀采样 但传入gsplat.project_gaussians_2d函数的时候，传入的第二个参数应该是协方差矩阵cholesky分解后的下三角矩阵。 这块代码我是理解错了吗，还是有啥简化和假设论文中没提到？ 期待大佬的回复，谢谢！
Hi, thanks for your work! I reproduced the training code based on GaussianSR, but the model weights I trained did not perform as well as expected.  Note: The left...
Hello, In the paper you reported results only with SwinIR as the backbone. I was wondering whether you also ran experiments with EDSR or RDN backbones. Methods such as LIIF,...
I use a random size image to upscale. size(w, h) = (441, 446). Scale factor = (2, 2) There are lots of black patterns on the result image.  ...
作者您好，有两个问题希望请教一下： （1） ”weighted_cholesky = para_ / 4; weighted_cholesky[:, 0] *= scale2; weighted_cholesky[:, 1] *= scale2; weighted_cholesky[:, 2] *= scale1”这几行代码的逻辑是什么？在正文中似乎并没有对应的解释。 （2）正文中写到"σ_x^2 , σ_y^2 , and ρσ_x_σy fall within the ranges of...
我认为这是一个非常优秀的工作，但是模型内部有些我不理解的问题，比如： ``` python window_size = 1 # Window size for Gaussian position adjustments pred = [] # List to store predictions bs, _, _, _ = feat.shape # Batch size...
