[AAAI 2021] Split then Refine: Stacked Attention-guided ResUNets for Blind Single Image Visible Watermark Removal
Home Page: https://arxiv.org/abs/2012.07007
🔭 I’m currently working on Image/Video Editing and Synthesis. please visit my website for further details.
⏰ Very busy days! I am trying my best to solve the issues!
📍 We are hiring research interns to publish high-quality research papers! See more information here.
☕️ You can buy me a coffee, if you find my project helpful.
I have cobuild a discord server for text-to-video generation, try the server here: .
Hi, I was trying to train the model on the Logo Dataset that you had provided, without resizing the Images.
Instead of Hard resize to 256x256 which distorts the aspect ratio, I decided to use random cropping however after a few epochs I'd run into CUDA error
Here is the error:
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [0,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [1,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [2,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [3,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [4,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [5,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [6,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [7,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [8,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [9,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [10,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [11,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [12,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [13,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [14,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [15,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [16,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [17,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [18,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [19,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [20,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [21,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [22,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [23,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [24,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [25,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [26,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [27,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [28,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [29,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [30,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [228,0,0], thread: [31,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [32,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [33,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [34,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [35,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [36,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [37,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [38,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [39,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [40,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [41,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [42,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [43,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [44,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [45,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [46,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [47,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [48,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [49,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [50,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [51,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [52,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [53,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [54,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [55,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [56,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [57,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [58,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [59,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [60,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [61,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [62,0,0] Assertion input_val >= zero && input_val <= one failed.
/pytorch/aten/src/ATen/native/cuda/Loss.cu:102: operator(): block: [197,0,0], thread: [63,0,0] Assertion input_val >= zero && input_val <= one failed.
Traceback (most recent call last):
File "/home/qblocks/shashank/Development/Oct_21/Watermark_removal/deep-blind-watermark-removal_patch/main.py", line 71, in
main(args)
File "/home/qblocks/shashank/Development/Oct_21/Watermark_removal/deep-blind-watermark-removal_patch/main.py", line 41, in main
Machine.train(epoch)
File "/home/qblocks/shashank/Development/Oct_21/Watermark_removal/deep-blind-watermark-removal_patch/scripts/machines/VX.py", line 129, in train
l2_loss,att_loss,wm_loss,style_loss,ssim_loss = self.loss(outputs[0],self.norm(target),outputs[1],mask,outputs[2],self.norm(wm))
File "/home/qblocks/shashank/Development/Oct_21/Watermark_removal/wm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
result = self.forward(*input, **kwargs)
File "/home/qblocks/shashank/Development/Oct_21/Watermark_removal/deep-blind-watermark-removal_patch/scripts/machines/VX.py", line 85, in forward
att_loss = self.attLoss(pred_ms, mask)
File "/home/qblocks/shashank/Development/Oct_21/Watermark_removal/wm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
result = self.forward(*input, **kwargs)
File "/home/qblocks/shashank/Development/Oct_21/Watermark_removal/wm/lib/python3.8/site-packages/torch/nn/modules/loss.py", line 530, in forward
return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)
File "/home/qblocks/shashank/Development/Oct_21/Watermark_removal/wm/lib/python3.8/site-packages/torch/nn/functional.py", line 2525, in binary_cross_entropy
return torch._C._nn.binary_cross_entropy(
RuntimeError: CUDA error: device-side assert triggered
Initially I thought that this might be due to errors in the input mask as it expects the values to be b/w 0 and 1. However, upon printing the values of Mask and pred_ms (prediction), it was found that the model prediction tensor was NaN.
mask tensor([[[[1., 1., 1., ..., 0., 0., 0.],
[1., 1., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.],
...,
[1., 1., 1., ..., 0., 0., 0.],
[1., 1., 1., ..., 0., 0., 0.],
[1., 1., 1., ..., 0., 0., 0.]]],
[[[0., 0., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.],
...,
[0., 0., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.]]],
[[[0., 0., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.],
...,
[0., 0., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.]]],
[[[0., 0., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.],
...,
[1., 0., 0., ..., 1., 1., 1.],
[1., 1., 0., ..., 1., 1., 0.],
[1., 1., 1., ..., 1., 0., 0.]]]], device='cuda:0')
pred_mask tensor([[[[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
...,
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan]]],
[[[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
...,
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan]]],
[[[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
...,
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan]]],
[[[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
...,
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan]]]], device='cuda:0',
grad_fn=<SigmoidBackward>)
Following is the code that I am using to Crop the patch:
from __future__ import print_function, absolute_import
import os
import csv
import numpy as np
import json
import random
import math
import matplotlib.pyplot as plt
from collections import namedtuple
from os import listdir
from os.path import isfile, join
import torch
# torch.manual_seed(17)
import torch.utils.data as data
from scripts.utils.osutils import *
from scripts.utils.imutils import *
from scripts.utils.transforms import *
import torchvision.transforms as transforms
from PIL import Image
from PIL import ImageEnhance
from PIL import ImageFilter
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True
import glob
class COCO(data.Dataset):
def __init__(self,train,config=None, sample=[],gan_norm=False):
self.train = []
self.anno = []
self.mask = []
self.wm = []
self.input_size = config.input_size
self.normalized_input = config.normalized_input
self.base_folder = config.base_dir
self.dataset = train+config.data
if config == None:
self.data_augumentation = False
else:
self.data_augumentation = config.data_augumentation
self.istrain = False if self.dataset.find('train') == -1 else True
self.sample = sample
self.gan_norm = gan_norm
file_paths2 = sorted(glob.glob(join(self.base_folder,'wm_DIV2K','full_*',self.dataset,'image/*')))
for fl2 in file_paths2:
file_name2 = fl2.split('/')[-1]
self.train.append(fl2)
self.mask.append(fl2.replace('/image/','/mask/'))
self.wm.append(fl2.replace('/image/','/wm/'))
self.anno.append(os.path.join(self.base_folder,'wm_DIV2K','natural',self.dataset,file_name2.split('-')[0]+'.'+file_name2.split('.')[-1]))
if len(self.sample) > 0:
self.train = [ self.train[i] for i in self.sample ]
self.mask = [ self.mask[i] for i in self.sample ]
self.anno = [ self.anno[i] for i in self.sample ]
self.trans = transforms.Compose([
transforms.ToTensor(),
])
print('total Dataset of '+self.dataset+' is : ', len(self.train))
def __getitem__(self, index):
img = Image.open(self.train[index]).convert('RGB')
mask = Image.open(self.mask[index]).convert('L')
anno = Image.open(self.anno[index]).convert('RGB')
wm = Image.open(self.wm[index]).convert('RGB')
W, H = img.size
if W < self.input_size or H < self.input_size:
img = img.resize((self.input_size, self.input_size))
mask = mask.resize((self.input_size, self.input_size))
anno = anno.resize((self.input_size, self.input_size))
wm = wm.resize((self.input_size, self.input_size))
i, j, h, w = transforms.RandomCrop.get_params(img, output_size=(self.input_size, self.input_size))
img = transforms.functional.crop(img,i,j,h,w)
mask = transforms.functional.crop(mask,i,j,h,w)
anno = transforms.functional.crop(anno,i,j,h,w)
wm = transforms.functional.crop(wm,i,j,h,w)
img = self.trans(img)
anno = self.trans(anno)
mask = self.trans(mask)
wm = self.trans(wm)
return {"image": img,
"target": anno,
"mask": mask,
"wm": wm,
"name": self.train[index].split('/')[-1],
"imgurl":self.train[index],
"maskurl":self.mask[index],
"targeturl":self.anno[index],
"wmurl":self.wm[index]
}
def __len__(self):
return len(self.train)
Any Help would be appreciated!!
how to train the model with my own data?
Do you have the training script?
Thx
Dear XiaoDong,
When will you be able to upload the pre-trained model on the grey image？
Look forward to your favourable reply.
Hi, I did not find mask and wm folders after unzip the Datasets/27kpng.zip. Could you help me?
ModuleNotFoundError: No module named 'scripts.models.vmu2',
and how to inference one nature image ?
1
i tried your evaluate.sh. and downloaded your datasets, 10kgray, 10kmid
input "bash examples/evaluate.sh"
but got "FileNotFoundError: [Errno 2] No such file or directory: '/home/dell/watermark/10kgray/masks/natural/COCO_val2014.png'
What's "COCO_val2014.png"?
would this model be available on videos ?
I have multiple GPUs and I would Like to train the model on them for faster training.
I see that you have already implemented MultiGPU training by using nn.DataParallel. There were some bugs in VX.py which were solved after i converted "self.model" to "self.model.module".
Yet after ensuring that i am using "CUDA_VISIBLE_DEVICES=0,1" i still see only GPU0 's memory to be filled and not GPU1's.
The model gives cuda out of memory if i try to use a input size >=512 with a batchsize of 12 or even 8.
Any idea why is it only consuming 1 of the 2 GPUs??
Thanks
Kindly share your supplementary paper which has the model architecture changes shared mentioned in your paper. Thanks
Excuse me , can I know how to evaluate the result of removing watermarks?
Do you have some metric like mAP or IoU? I just wonder how to evaluate the results of removing watermarks.
Can I get some evaluating metric or something else?
Thx!
Hi, thank you for sharing your work.
I want to run your network on my data, But when I decompressed 27kpng_model_best.pth.tar, it says the file is corrupted, could you please upload it again？
Hi, I did not find mask and wm folders after unzip the Datasets/27kpng.zip. Could you help me?
code as follow:
from PIL import Image
img=Image.open('XX.jpg').resize((256,256))
image=np.array(img)
ih=torch.tensor(np.transpose(image,(2,0,1)))/255.0
ih=ih.unsqueeze(0)
imoutput,immask,imwatermark= model(ih.cuda())
imcoarser,imrefine,imwatermark = imoutput[1]immask + im(1-immask),imoutput[0]immask + im(1-immask),imwatermark*immask
torchvision.utils.save_image(imcoarser,'imrefine.jpg')
Hi Vinthony,
I saw that you are resizing the input image to 256x256. Is there any support for variable size inputs?
Hi, can I try it on watermarked image which I do not have mask. If so could you provide example on colab file removing watermark from uploaded image or from a test folder?
Thanks for your excellent work! However, I found some corrupted images in the dataset. In 10kgray, when reading val_images/image/COCO_val2014_000000087481-Kaporal_Jeans_Logo-185.png, train_images/image/COCO_val2014_000000260525-Breyers_Logo-163.png, train_images/image/COCO_val2014_000000510651-University_Of_Oxford_Logo_Text-179.png using imageio.v2.imread, an error image file is truncated occurs. In fact the latter two images do not contain the watermark and are thus not usable.
Thanks for your impressive work.
I have some question about dataset in the onedrive link.
I guess:
10kmid.zip is LOGO-L
10khigh.zip is LOGO-H
10kgray.zip is LOGO-Gray
27kpng.zip is LOGO-30k
I wonder if this mapping is correct because there is some difference in naming.
Looking forward to your reply.
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.