EleGANt icon indicating copy to clipboard operation
EleGANt copied to clipboard

PyTorch code for "EleGANt: Exquisite and Locally Editable GAN for Makeup Transfer" (ECCV 2022)

Results 12 EleGANt issues
Sort by recently updated
recently updated
newest added

老师您好,用自己的图片运行demo,输出的结果是被裁减后的图片,请问如何修改可以输出没有被裁减的图片呢?

您好,感谢您的elegant工作, 想询问下我在调整分辨率到512后训练batch_size=1(A100)需要52gb的显存,2.5小时每epoch 在训练batch_zize=2时显存爆掉 然后希望使用多卡训练,该如何使用呢?

您好: 不好意思想請問作者face detection設置up ratio, down ratio跟width ratio的用意? 以及這個參數數值是如何訂的呢? 因為我將其他人臉資料集作為測試集時,發現經過這些ratio反而會對face detection產生反效果。例如本來抓到的bounding box是正常的,但經過ratio校正後bounding box偵測到的人臉區塊反而會偏移許多,如下圖所示人臉的部分反而抓的不多。 因此想詢問作者這些參數該如何設置,或是可以去除這些參數設置嗎? 目前我是拿video的人臉資料集套github方法。首先是遇到這個問題,但不知道該怎麼設置參數比較好(或是不知道是否可以拔除這個部份),也不知道參數設置的用意。 ![示意圖](https://user-images.githubusercontent.com/44439517/208253320-85059a93-55c8-4cca-a10e-25bf3386b269.png) 謝謝作者的回覆

How can I finetune the model? I am trying to load the model and start training but the seg folder is missing in my dataset. How can I create the...

Hi, thanks for great work, I have a question as follows: how to get global shade control result as Fig 1 shows in the paper? which shade can be adjust...

![image](https://user-images.githubusercontent.com/41255670/229009657-403b6383-a5b3-4821-ae20-0f41afd23f47.png)

Can the author provide a training dataset? especially in numpy format

Thank you for your outstanding work. How can I obtain images generated by PGT ![image](https://user-images.githubusercontent.com/127761312/226548803-c7169dc2-8fbd-4c92-92d5-0a22b5d32f5f.png)