GFPGAN 1024
License: MIT License
I am Leslie from China, nice to meet you all!
your instructions say:
...
7. get discriminator here which is transformed from original stylegan2
...
I can't download the file: d.pth (404 error)
fix the link
Thank you for your work; I've found the results to be significantly better than those from GFPGAN 1.4. I am attempting to reproduce your results but have noticed some discrepancies.
In config.py, should self.crop_components = False actually be set to self.crop_components = True?
Regarding the instruction "train until you think it is okay," how can I be certain that it is indeed okay?
I ask you to add your model to the replicate.com site so that other users can use your model.
I cannot use your provided website aliyun.com in Chinese.
Thank you for your great work. The current image input does not support batch_size （infer）, can you support it?
感谢您杰出的工作，当前图片输入不支持 batch_size （推理），你能支持它吗？
Hi please help me get link download list pth file:
self.scratch_d_path = 'pretrained_models/d.pth'
self.scratch_left_eye_path = 'pretrained_models/GFPGANv1_net_d_left_eye.pth'
self.scratch_right_eye_path = 'pretrained_models/GFPGANv1_net_d_right_eye.pth'
self.scratch_mouth_path = 'pretrained_models/GFPGANv1_net_d_mouth.pth'
thank you
Hi, I have been trying to train ffhqr at 1024 x 1024 resolution. Here are the changes in my config file. I am using ffhq as low-quality images.
class Params:
def __init__(self):
self.name = 'GFPGAN'
self.mode = 'encoder'
self.pretrain_path = 'GFPGAN-1024/checkpoint/GFPGAN/decoder/001-00027000.pth'
self.scratch_gan_path = 'pretrained_models/GFPGANv1.4.pth'
self.scratch_d_path = 'pretrained_models/GFPGANv1_net_d.pth'
self.scratch_left_eye_path = 'pretrained_models/GFPGANv1_net_d_left_eye.pth'
self.scratch_right_eye_path = 'pretrained_models/GFPGANv1_net_d_right_eye.pth'
self.scratch_mouth_path = 'pretrained_models/GFPGANv1_net_d_mouth.pth'
self.id_model = 'pretrained_models/arcface_resnet18.pth'
self.img_root = "output/train/ffhqr"
self.train_hq_root = "output/train/ffhqr"
self.train_lq_root = 'output/train/ffhq'
self.train_lmk_base = '' # lmk info
self.val_lmk_base = ''
self.val_lq_root = 'output/val/ffhq'
self.val_hq_root = 'output/val/ffhqr'
self.g_lr = 1e-3I have commented out lines 210 to 228 in dataloader/GFPLoader.py since I provide low-quality images, I did not need to produce them.
# ------------------------ generate lq image ------------------------ #
# blur
'''
if not self.eval:
kernel = degradations.random_mixed_kernels(
self.kernel_list,
self.kernel_prob,
self.blur_kernel_size,
self.blur_sigma,
self.blur_sigma, [-math.pi, math.pi],
noise_range=None)
img_lq = cv2.filter2D(img_lq, -1, kernel)
# downsample
scale = np.random.uniform(self.downsample_range[0], self.downsample_range[1])
img_lq = cv2.resize(img_lq, (int(w // scale), int(h // scale)), interpolation=cv2.INTER_LINEAR)
# # noise
if self.noise_range is not None:
img_lq = degradations.random_add_gaussian_noise(img_lq, self.noise_range)
# jpeg compression
if self.jpeg_range is not None:
img_lq = degradations.random_add_jpg_compression(img_lq, self.jpeg_range)
'''
# resize to original size
img_lq = cv2.resize(img_lq, (512, 512), interpolation=cv2.INTER_LINEAR)
img_hq = cv2.resize(img_hq, (1024, 1024))Why img_lq is resized to 512x512 because I want to train on ffhq @ 1024x1024 to produce ffhqr 1024x1024?
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
Django
The Web framework for perfectionists with deadlines.
Laravel
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft
Open source projects and samples from Microsoft.
Google
Google ❤️ Open Source for everyone.
Alibaba
Alibaba Open Source for everyone
D3
Data-Driven Documents codes.
Tencent
China tencent open source team.