Tang, Wenyi
Tang, Wenyi
> 我觉得超分是不是可以用于视频压缩传输 > 1.压缩算法是已知的,可以轻松训练得到模糊核 > 2.带宽有限,用LR的方式可以更容易传输 > 3.在播放端使用超分将LR还原为HR > 请各位大佬指正 前提条件很难成立:**为什么不直接降低码率传输?**事实上所有的视频推流服务都是这么做的。**同码率**下高分辨率(HR)的视频质量远远高于低分辨率(LR),也远高于超分后的SR图像。
Hi @mesw, I very much appreciate your help! First of all, you can start within `VSRTorch`, by simply writing tdan.py into `VSRTorch/Models/`; Then link your model by adding an entry...
@Easoncyx Thanks, I will take a look
Hi @Iffa-Meah The ID is not from tensor name, instead it's from tensor source's **compile option**, I don't think we can create a specific handler for that.
Hi @Iffa-Meah, please see model files at https://github.com/LoSealL/openvino/releases/tag/test On my PC (_i7-9750H with UHD 630 GEN9 24EU_), it will dump **20** .cl_cache, but only **19** .cl files, because two of...
There's no upsample module in **VDSR**. It convs on bicubically upsampled images.
> $global:GitPromptSettings.AnsiConsole = $false Same issue and this works for me
这个也太依赖electron的版本控制了,一个不兼容就是灾难
g2o 版本问题吧,如果你用最新的g2o的话需要稍微修改一下例子中的代码,接口参数有些改变,g2o有些接口用std::unique_ptr取代了裸指针。
@Junkai-Liu 年代久远有点记忆迷糊了,你的bin指的是这个example的执行文件? 把你改后的代码贴一下 参考一下我的[fork](https://github.com/LoSealL/g2o_ba_example),这是一年前的g2o版本,不保证对现在的仍然有效