{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# Introduction\n\nThis kernel contains the inference code of the submission_experts team for the 2021 HuBMAP competition (version 1). Some additional insights are provided at the end of the notebook.\n\nThe goal of the competition was to provide with automated glomeruli segmentation models for WSI of human kidneys. In this competition, these WSI can be divided into two categories: those containing fresh-frozen (FFPE) tissues and those containing formalin-fixed paraffin-embedded tissues. Obvious visual differences can be noticed in both types of WSI: the structure of FFPE tissues is better preserved than fresh-frozen ones. Therefore, we initially tried to apply different models with respect to the type of WSI that we were processing. That approach was giving us very good results on the public leaderboard and on cross-validation before the change of test data, but unfortunately for some reason, we could not see that improvement afterwards. After discussion, we decided not to follow that approach in our final submissions.","metadata":{}},{"cell_type":"code","source":"!mkdir -p /tmp/pip/cache/\n!cp ../input/segmentationmodelspytorch/segmentation_models/efficientnet_pytorch-0.6.3.xyz /tmp/pip/cache/efficientnet_pytorch-0.6.3.tar.gz\n!cp ../input/segmentationmodelspytorch/segmentation_models/pretrainedmodels-0.7.4.xyz /tmp/pip/cache/pretrainedmodels-0.7.4.tar.gz\n!cp ../input/segmentationmodelspytorch/segmentation_models/segmentation-models-pytorch-0.1.2.xyz /tmp/pip/cache/segmentation_models_pytorch-0.1.2.tar.gz\n!cp ../input/segmentationmodelspytorch/segmentation_models/timm-0.1.20-py3-none-any.whl /tmp/pip/cache/\n!cp ../input/segmentationmodelspytorch/segmentation_models/timm-0.2.1-py3-none-any.whl /tmp/pip/cache/\n!pip install --no-index --find-links /tmp/pip/cache/ efficientnet-pytorch\n!pip install --no-index --find-links /tmp/pip/cache/ segmentation-models-pytorch\n\n# Imports\nimport sys\nsys.path.insert(0, \"../input/resnest/\")\nimport resnest.torch as resnest_torch\n\nimport numpy as np\nimport pandas as pd\nimport torch\nimport torch.nn as nn\nimport torch.nn.functional as F\nimport matplotlib.pyplot as plt\nfrom PIL import Image\nimport random\nimport rasterio\nimport tempfile\nimport cv2\nimport os\nimport shutil\nimport gc\nfrom tqdm.notebook import tqdm\nimport time\nimport skimage.measure\n\nfrom torch.utils.data import Dataset, DataLoader\nfrom torchvision.models.resnet import ResNet, Bottleneck\nimport segmentation_models_pytorch as smp\n\nfrom fastai.vision.all import PixelShuffle_ICNR, ConvLayer # TODO: remove\n\nimport warnings\nwarnings.filterwarnings(\"ignore\")","metadata":{"_kg_hide-input":false,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Training Tricks\n\n## Disclamer\n\nWe used NO hand-labelling and manual annotation since in our vision it is not consistent with the initial goal and objective of the competition.\n\n## Color spaces augmentation trick:\n\nFor the training we found and decided to use various Color Space Augmentations in combination with custom stochastic kernel - this is one of the most important things that allowed our models to stay robust and segment glomeruli despite the data source, type of the images (FFPE/fresh-frozen/even others) and \"color-related\" variations. \n\nOn the images below one could see how the naturally looked tiles are transformed being augmented with this method. You can see normal tiles with the cutmix augs, color spaced without cutmix, and color spaced with cutmix.\n\n\n**Tiles with CutMix augmentation:**\n\n![image.png](attachment:c3ebe340-3998-4698-a504-39214ecbfa4f.png)\n\n**Tiles with Color Space augmentation:**\n\n![image.png](attachment:9bb3d996-0be8-44ec-9a9a-05d8135be713.png)\n\n**Tiles with both augmentations:**\n\n![image.png](attachment:af404411-8d4e-4df3-9bb8-311a51da9d32.png)\n\nThis methodology demonstrated its effectiveness and power during both stages of the competition, but mostly with the first set of the data (maybe due to the different test sets, since we dont know yet results on the private test data).\n\nWe also share the code as we implemented this idea.\n\n\n```\nimport cv2\n\ncspaces = [cv2.COLOR_BGR2HLS,\n\n            cv2.COLOR_BGR2HSV,\n\n            cv2.COLOR_BGR2LAB,\n\n            cv2.COLOR_BGR2LUV,\n\n            cv2.COLOR_BGR2Lab,\n\n            cv2.COLOR_BGR2Luv,\n\n            cv2.COLOR_BGR2RGB,\n\n            cv2.COLOR_BGR2XYZ,\n\n            cv2.COLOR_BGR2YUV,\n\n            cv2.COLOR_RGB2HLS,\n\n            cv2.COLOR_RGB2HSV,\n\n            cv2.COLOR_RGB2LAB,\n\n            cv2.COLOR_RGB2LUV,\n\n            cv2.COLOR_RGB2Lab,\n\n            cv2.COLOR_RGB2Luv,\n\n            cv2.COLOR_RGB2BGR,\n\n            cv2.COLOR_RGB2XYZ,\n\n            cv2.COLOR_RGB2YUV]\n```\n\n\nIn the batch sampling part:\n \n\n```\nif self.train and random.random() > 1/len(cspaces):\n\n    cspace = random.choice(cspaces)\n\n    img = cv2.cvtColor(img, cspace)\n\nif self.train:\n\n    stoch = np.random.rand(3,3)\n\n    K = stoch.sum(0, keepdims=True)\n\n    if random.random() > 0.7:\n\n        K=np.exp(K)/K\n\n    elif random.random() > 0.25:\n\n        K=np.exp(K)/(np.sqrt(K))\n\n    stoch = stoch/K\n\n    img = np.einsum(\"ijk,kl->ijl\", img, stoch)\n```\n\nThis method also allowed to obtain much more focused predictions - we observed very high “soft dice” metrics even without proper binarization threshold. For example improvement is from 0.6-0.65 in the standard pipeline up to 0.87-0.89 Dice, depending on the fold.\n\nAdditionally we observed interesting behaviour such as robust segmentation even of some suspicious glomeruli in the d48*** sample from public test set.\n\n## Multi-scale input tiles\n\nWe used multiple image reduction rates (3, 4, 5, 6 and 8) in order to create the tiles from original images. Additionally we used only “non-empty” tiles of the bigger reduction rates to balance more the training dataset. Such approach naturally increases the training dataset and enriches variation of glomeruli size and its proportion in the considered tile of fixed size.\n\nThis also gave us significant boost both in CV and LB.\n\n## Pseudo-labelling\n\nPublic test set pseudo-labeled with own models and included in the training procedure\n\n## Short training pipeline\n\nUnet with resnext50_32x4d/resnext101_32x16d/efficientnet-b7/b5 backbones, ASPP and FPN modules + complex loss (combination of FocalLoss + DiceLoss + SymmetricLovasz) + OneCycleLR + Heavy Augmentations & CutMix & ColorSpaced augs + 5-fold patient-based CV.","metadata":{},"attachments":{"c3ebe340-3998-4698-a504-39214ecbfa4f.png":{"image/png":"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"},"9bb3d996-0be8-44ec-9a9a-05d8135be713.png":{"image/png":"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"},"af404411-8d4e-4df3-9bb8-311a51da9d32.png":{"image/png":"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"}}},{"cell_type":"markdown","source":"# Inference\n\nIn this section, we describe how we make predictions.\n\n## Inference in a nutshell\n\n- We make an inference on the bigger patch size than was used for training. Before the data update the best one was of 1024 and after - 512 while the training is performed on 256x256.\n\n- We used overlapping step of 0.5xtile_size in both X and Y directions to cut the patches from test images. Locally we used smaller step of 0.33 that wasn't used because of the inference time constraints.\n\n- We also used classical TTA based on transformations of the D4 group, but we were limited by the kernel inference time.\n\n- In order to handle the WSI images size we used memmap to load the images, and also used temporary saving of processed patches to perform the ensembling.\n\n- Very important point that we wanted to highlight, that we were able to split all test images into two groups with respect to the ffpe/ff type using the \"PhysicalSizeY\" from the tiff.pages.description. This information was crucial before the data update and allowed us to get different models that worked for each of the groups. However, by some reasons (we did not discover which) this stopped working after the data update (at least for the public LB). But it still demonstrated higher CV when models were trained independently for each group.\n\n- This version makes a two-pass segmentation: a first pass aims at finding where glomeruli are (detection threshold is very low) and a second pass aims at refining segmentations by applying our models on tiles centered on glomeruli found during the first pass.\n\n- Final submissions are ensembles of models trained with and without Color Space Augmenatations and Cutmix in order to increase the robustness.\n\n## Parameters\n\nHere are the parameters that we use for inference.","metadata":{}},{"cell_type":"code","source":"# PARAMETERS\n\n# Printing parameters\nVERBOSE = True\n\n# Data processing\nDATA_DIR = '../input/hubmap-kidney-segmentation/test' # Input data directory\nREDUCTION = 3 # Reduce the original images by x times\nTILE_SZ = 768 # Size of tiles on which inference is done\nTILE_SZ_2nd = 384\n# https://www.kaggle.com/iafoss/256x256-images\n#MEAN = np.array([0.65459856,0.48386562,0.69428385])\n#STD = np.array([0.15167958,0.23584107,0.13146145])\nMEAN = np.array([0.63482309,0.47376275,0.67814029])\nSTD = np.array([0.17405236,0.23305763,0.1585981])\n\n# Models for first pass\nMODELS_PATHS_1st = [f'../input/ens-red345/model_ux50_{i}.pth' for i in range(2)]\n\n# Models for second pass\nMODELS_PATHS_2nd = [f'../input/ens-red345/model_effb5_{i}.pth' for i in range(2)]\nMODELS_PATHS_2nd += [f'../input/ens-red345/model_effb7_{i}.pth' for i in [1,3]]\nMODELS_PATHS_2nd += [f'../input/ret-r101-multi3468-lf/model_{i}.pth' for i in [0,2]]\n#MODELS_PATHS_2nd += [f'../input/ens-red345/model_rnst200_{i}.pth' for i in range(1)]\n\n# Tiles selection\n# https://www.kaggle.com/iafoss/256x256-images\nS_TH = 40 # Saturation blancking threshold\nP_TH = 200*TILE_SZ//256 # Threshold for the minimum number of pixels\n\n# Size of center check box\nCHECK_SZ = 256\n\n# Inference\nPUBLIC_ONLY = False # Make predictions only on public LB\nX_OVERLAP = [0., 0.5] # Overlap between tiles during prediction (X axis)\nY_OVERLAP = [0., 0.5] # Overlap between tiles during prediction (Y axis)\nCUSTOM_RED = 1 # Reduction for two types of models\nTH_1st = 0.1 # Threshold for first pass\nTH_2nd = 0.3 # Threshold for second pass\nN_BINS = 255 # Number of bins when saving mask tiles\nBATCH_SIZE = 2\nNUM_WORKERS = 2\nHALF_PRECISION = False\nTTA_FLIPS = [[-1], [-2], [-2, -1]]\nROT_TTA_FLIPS = [0, 1]\n\n# Final prediction\nMASK_SZ = 4096 # Size of saved mask tiles","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n\nif os.path.exists('tmp'):\n    if VERBOSE:\n        print(\"Removing 'tmp' directory\")\n    shutil.rmtree('tmp')","metadata":{"_kg_hide-output":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### Models\n\nThe models that we can use for inference are defined below.","metadata":{}},{"cell_type":"code","source":"class FPN(nn.Module):\n    def __init__(self, input_channels:list, output_channels:list):\n        super().__init__()\n        self.convs = nn.ModuleList(\n            [nn.Sequential(nn.Conv2d(in_ch, out_ch*2, kernel_size=3, padding=1),\n             nn.ReLU(inplace=True), nn.BatchNorm2d(out_ch*2),\n             nn.Conv2d(out_ch*2, out_ch, kernel_size=3, padding=1))\n            for in_ch, out_ch in zip(input_channels, output_channels)])\n        \n    def forward(self, xs:list, last_layer):\n        hcs = [F.interpolate(c(x),scale_factor=2**(len(self.convs)-i),mode='bilinear') \n               for i,(c,x) in enumerate(zip(self.convs, xs))]\n        hcs.append(last_layer)\n        return torch.cat(hcs, dim=1)\n\nclass UnetBlock(nn.Module):\n    def __init__(self, up_in_c:int, x_in_c:int, nf:int=None, blur:bool=False,\n                 self_attention:bool=False, **kwargs):\n        super().__init__()\n        self.shuf = PixelShuffle_ICNR(up_in_c, up_in_c//2, blur=blur, **kwargs)\n        self.bn = nn.BatchNorm2d(x_in_c)\n        ni = up_in_c//2 + x_in_c\n        nf = nf if nf is not None else max(up_in_c//2,32)\n        self.conv1 = ConvLayer(ni, nf, norm_type=None, **kwargs)\n        self.conv2 = ConvLayer(nf, nf, norm_type=None,\n            xtra=SelfAttention(nf) if self_attention else None, **kwargs)\n        self.relu = nn.ReLU(inplace=True)\n\n    def forward(self, up_in:torch.Tensor, left_in:torch.Tensor) -> torch.Tensor:\n        s = left_in\n        up_out = self.shuf(up_in)\n        cat_x = self.relu(torch.cat([up_out, self.bn(s)], dim=1))\n        return self.conv2(self.conv1(cat_x))\n        \nclass _ASPPModule(nn.Module):\n    def __init__(self, inplanes, planes, kernel_size, padding, dilation, groups=1):\n        super().__init__()\n        self.atrous_conv = nn.Conv2d(inplanes, planes, kernel_size=kernel_size,\n                stride=1, padding=padding, dilation=dilation, bias=False, groups=groups)\n        self.bn = nn.BatchNorm2d(planes)\n        self.relu = nn.ReLU()\n\n        self._init_weight()\n\n    def forward(self, x):\n        x = self.atrous_conv(x)\n        x = self.bn(x)\n\n        return self.relu(x)\n\n    def _init_weight(self):\n        for m in self.modules():\n            if isinstance(m, nn.Conv2d):\n                torch.nn.init.kaiming_normal_(m.weight)\n            elif isinstance(m, nn.BatchNorm2d):\n                m.weight.data.fill_(1)\n                m.bias.data.zero_()\n\nclass ASPP(nn.Module):\n    def __init__(self, inplanes=512, mid_c=256, dilations=[6, 12, 18, 24], out_c=None):\n        super().__init__()\n        self.aspps = [_ASPPModule(inplanes, mid_c, 1, padding=0, dilation=1)] + \\\n            [_ASPPModule(inplanes, mid_c, 3, padding=d, dilation=d,groups=4) for d in dilations]\n        self.aspps = nn.ModuleList(self.aspps)\n        self.global_pool = nn.Sequential(nn.AdaptiveMaxPool2d((1, 1)),\n                        nn.Conv2d(inplanes, mid_c, 1, stride=1, bias=False),\n                        nn.BatchNorm2d(mid_c), nn.ReLU())\n        out_c = out_c if out_c is not None else mid_c\n        self.out_conv = nn.Sequential(nn.Conv2d(mid_c*(2+len(dilations)), out_c, 1, bias=False),\n                                    nn.BatchNorm2d(out_c), nn.ReLU(inplace=True))\n        self.conv1 = nn.Conv2d(mid_c*(2+len(dilations)), out_c, 1, bias=False)\n        self._init_weight()\n\n    def forward(self, x):\n        x0 = self.global_pool(x)\n        xs = [aspp(x) for aspp in self.aspps]\n        x0 = F.interpolate(x0, size=xs[0].size()[2:], mode='bilinear', align_corners=True)\n        x = torch.cat([x0] + xs, dim=1)\n        return self.out_conv(x)\n    \n    def _init_weight(self):\n        for m in self.modules():\n            if isinstance(m, nn.Conv2d):\n                torch.nn.init.kaiming_normal_(m.weight)\n            elif isinstance(m, nn.BatchNorm2d):\n                m.weight.data.fill_(1)\n                m.bias.data.zero_()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class UneXt50(nn.Module):\n    def __init__(self, stride=1, **kwargs):\n        super().__init__()\n        #encoder\n        m = ResNet(Bottleneck, [3, 4, 6, 3], groups=32, width_per_group=4)\n        #m = torch.hub.load('facebookresearch/semi-supervised-ImageNet1K-models',\n        #                   'resnext50_32x4d_ssl')\n        self.enc0 = nn.Sequential(m.conv1, m.bn1, nn.ReLU(inplace=True))\n        self.enc1 = nn.Sequential(nn.MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1),\n                            m.layer1) #256\n        self.enc2 = m.layer2 #512\n        self.enc3 = m.layer3 #1024\n        self.enc4 = m.layer4 #2048\n        #aspp with customized dilatations\n        self.aspp = ASPP(2048,256,out_c=512,dilations=[stride*1,stride*2,stride*3,stride*4])\n        self.drop_aspp = nn.Dropout2d(0.5)\n        #decoder\n        self.dec4 = UnetBlock(512,1024,256)\n        self.dec3 = UnetBlock(256,512,128)\n        self.dec2 = UnetBlock(128,256,64)\n        self.dec1 = UnetBlock(64,64,32)\n        self.fpn = FPN([512,256,128,64],[16]*4)\n        self.drop = nn.Dropout2d(0.1)\n        self.final_conv = ConvLayer(32+16*4, 1, ks=1, norm_type=None, act_cls=None)\n        \n    def forward(self, x):\n        enc0 = self.enc0(x)\n        enc1 = self.enc1(enc0)\n        enc2 = self.enc2(enc1)\n        enc3 = self.enc3(enc2)\n        enc4 = self.enc4(enc3)\n        enc5 = self.aspp(enc4)\n        dec3 = self.dec4(self.drop_aspp(enc5),enc3)\n        dec2 = self.dec3(dec3,enc2)\n        dec1 = self.dec2(dec2,enc1)\n        dec0 = self.dec1(dec1,enc0)\n        x = self.fpn([enc5, dec3, dec2, dec1], dec0)\n        x = self.final_conv(self.drop(x))\n        x = F.interpolate(x,scale_factor=2,mode='bilinear')\n        return x\n    \n    \n    \nclass UneXt101(nn.Module):\n    def __init__(self, stride=1, **kwargs):\n        super().__init__()\n        #encoder\n        m = ResNet(Bottleneck, [3, 4, 23, 3], groups=32, width_per_group=16)\n\n        self.enc0 = nn.Sequential(m.conv1, m.bn1, nn.ReLU(inplace=True))\n        self.enc1 = nn.Sequential(nn.MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1),\n                            m.layer1) #256\n        self.enc2 = m.layer2 #512\n        self.enc3 = m.layer3 #1024\n        self.enc4 = m.layer4 #2048\n        #aspp with customized dilatations\n        self.aspp = ASPP(2048,256,out_c=512,dilations=[stride*1,stride*2,stride*3,stride*4])\n        self.drop_aspp = nn.Dropout2d(0.5)\n        #decoder\n        self.dec4 = UnetBlock(512,1024,256)\n        self.dec3 = UnetBlock(256,512,128)\n        self.dec2 = UnetBlock(128,256,64)\n        self.dec1 = UnetBlock(64,64,32)\n        self.fpn = FPN([512,256,128,64],[16]*4)\n        self.drop = nn.Dropout2d(0.1)\n        self.final_conv = ConvLayer(32+16*4, 1, ks=1, norm_type=None, act_cls=None)\n        \n    def forward(self, x):\n        enc0 = self.enc0(x)\n        enc1 = self.enc1(enc0)\n        enc2 = self.enc2(enc1)\n        enc3 = self.enc3(enc2)\n        enc4 = self.enc4(enc3)\n        enc5 = self.aspp(enc4)\n        dec3 = self.dec4(self.drop_aspp(enc5),enc3)\n        dec2 = self.dec3(dec3,enc2)\n        dec1 = self.dec2(dec2,enc1)\n        dec0 = self.dec1(dec1,enc0)\n        x = self.fpn([enc5, dec3, dec2, dec1], dec0)\n        x = self.final_conv(self.drop(x))\n        x = F.interpolate(x,scale_factor=2,mode='bilinear')\n        return x\n    \nclass Unet50(nn.Module):\n    def __init__(self, stride=1, **kwargs):\n        super().__init__()\n        #encoder\n#         m = torch.hub.load('facebookresearch/semi-supervised-ImageNet1K-models',\n#                            'resnet50_swsl')\n        m = ResNet(Bottleneck, [3, 4, 6, 3])\n        self.enc0 = nn.Sequential(m.conv1, m.bn1, nn.ReLU(inplace=True))\n        self.enc1 = nn.Sequential(nn.MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1),\n                            m.layer1) #256\n        self.enc2 = m.layer2 #512\n        self.enc3 = m.layer3 #1024\n        self.enc4 = m.layer4 #2048\n        #aspp with customized dilatations\n        self.aspp = ASPP(2048,256,out_c=512,dilations=[stride*1,stride*2,stride*3,stride*4])\n        self.drop_aspp = nn.Dropout2d(0.5)\n        #decoder\n        self.dec4 = UnetBlock(512,1024,256)\n        self.dec3 = UnetBlock(256,512,128)\n        self.dec2 = UnetBlock(128,256,64)\n        self.dec1 = UnetBlock(64,64,32)\n        self.fpn = FPN([512,256,128,64],[16]*4)\n        self.drop = nn.Dropout2d(0.1)\n        self.final_conv = ConvLayer(32+16*4, 1, ks=1, norm_type=None, act_cls=None)\n        \n    def forward(self, x):\n        enc0 = self.enc0(x)\n        enc1 = self.enc1(enc0)\n        enc2 = self.enc2(enc1)\n        enc3 = self.enc3(enc2)\n        enc4 = self.enc4(enc3)\n        enc5 = self.aspp(enc4)\n        dec3 = self.dec4(self.drop_aspp(enc5),enc3)\n        dec2 = self.dec3(dec3,enc2)\n        dec1 = self.dec2(dec2,enc1)\n        dec0 = self.dec1(dec1,enc0)\n        x = self.fpn([enc5, dec3, dec2, dec1], dec0)\n        x = self.final_conv(self.drop(x))\n        x = F.interpolate(x,scale_factor=2,mode='bilinear')\n        return x\n\nclass UneSt200(nn.Module):\n    def __init__(self, stride=1, **kwargs):\n        super().__init__()\n        #encoder\n        m = getattr(resnest_torch, 'resnest200')(pretrained=False)\n    \n        self.enc0 = nn.Sequential(m.conv1, m.bn1, nn.ReLU(inplace=True))\n        self.enc1 = nn.Sequential(m.maxpool,\n                            m.layer1) #256\n        self.enc2 = m.layer2 #512\n        self.enc3 = m.layer3 #1024\n        self.enc4 = m.layer4 #2048\n        #aspp with customized dilatations\n        self.aspp = ASPP(2048,256,out_c=512,dilations=[stride*1,stride*2,stride*3,stride*4])\n        self.drop_aspp = nn.Dropout2d(0.3)\n        #decoder\n        self.dec4 = UnetBlock(512,1024,256)\n        self.dec3 = UnetBlock(256,512,128)\n        self.dec2 = UnetBlock(128,256,64)\n        self.dec1 = UnetBlock(64,128,32)\n        self.fpn = FPN([512,256,128,64],[16]*4)\n        self.drop = nn.Dropout2d(0.2)\n        self.final_conv = ConvLayer(32+16*4, 1, ks=1, norm_type=None, act_cls=None)\n        \n    def forward(self, x):\n        enc0 = self.enc0(x)\n        enc1 = self.enc1(enc0)\n        enc2 = self.enc2(enc1)\n        enc3 = self.enc3(enc2)\n        enc4 = self.enc4(enc3)\n        enc5 = self.aspp(enc4)\n        dec3 = self.dec4(self.drop_aspp(enc5),enc3)\n        dec2 = self.dec3(dec3,enc2)\n        dec1 = self.dec2(dec2,enc1)\n        dec0 = self.dec1(dec1,enc0)\n        x = self.fpn([enc5, dec3, dec2, dec1], dec0)\n        x = self.final_conv(self.drop(x))\n        x = F.interpolate(x,scale_factor=2,mode='bilinear')\n        return x\n\ndef Effb7Unet():\n    return smp.Unet(encoder_name='efficientnet-b7', classes=1, activation=None, encoder_weights=None)\n\ndef Effb5Unet():\n    return smp.Unet(encoder_name='efficientnet-b5', classes=1, activation=None, encoder_weights=None)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"We import them as follows.","metadata":{}},{"cell_type":"code","source":"# Model name to class\nmodelname2class = {\"ux50\": UneXt50,\n                   \"0.pth\": UneXt101,\n                   \"2.pth\": UneXt101,\n                   \"effb5\": Effb5Unet,\n                   \"effb7\": Effb7Unet,\n                   \"rnst200\": UneSt200}\n\n# Import models\nMODELS_1st = []\nfor path in MODELS_PATHS_1st:\n    state_dict = torch.load(path,map_location=torch.device('cpu'))\n    model_name = path.split(\"_\")[1]\n    model = modelname2class[model_name]()\n    model.load_state_dict(state_dict)\n    model.float()\n    model.eval()\n    model.to(device)\n    MODELS_1st.append(model)\n    del state_dict\n\nMODELS_2nd = []\nfor path in MODELS_PATHS_2nd:\n    state_dict = torch.load(path,map_location=torch.device('cpu'))\n    model_name = path.split(\"_\")[1]\n    model = modelname2class[model_name]()\n    model.load_state_dict(state_dict)\n    model.float()\n    model.eval()\n    model.to(device)\n    MODELS_2nd.append(model)\n    del state_dict","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Detailed Description of our Inference Approach\n\nIn this section, we describe our approach (version 1). The second version, using a two-pass segmentation, is described in a separate notebook.\n\n## Data Preprocessing\n\nData preprocessing consists in downsampling images. We also initially wanted to find to which group (fresh-frozen vs FFPE) they belonged to, but as we said before, we finally did not follow that approach. We copy our group-finding code for information.\n\n### Loading and Downsampling Whole-Slide Images\n\nThe main issue with WSI is that they are very big. Here is how we load an image, using NumPy MemMap.","metadata":{}},{"cell_type":"code","source":"def read_tiff(filename):\n    img = rasterio.open(filename)\n    W, H = img.shape\n    tmp = np.memmap(tempfile.TemporaryFile(), shape=(W, H, 3),\n                          dtype=np.uint8)\n    if len(img.subdatasets) == 3:\n        for i in range(3):\n            tmp[:,:,i] = rasterio.open(img.subdatasets[i]).read(1)\n    else:\n        for i in range(3):\n            tmp[:,:,i] = img.read(i+1)\n    return tmp\n\ndef load_image(filename):\n    img_id = filename.split(\"/\")[-1].split(\".\")[0]\n    img = read_tiff(filename)\n    if VERBOSE:\n        print(\"Initial size of %s:\" %(img_id,), img.shape)\n    return img","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Dealing with such big images is not convenient, we found that we should downsample data for better performance. After loading the image using the `load_image` function above, we apply the `_tile_resize_save` function below.","metadata":{}},{"cell_type":"code","source":"def _tile_resize_save(img, img_id, tile_sz, reduce=1):\n    \"\"\"\n    Divide WSI into small tiles, resize them and save them locally.\n    \"\"\"\n    x = 0\n    while x < img.shape[0]:\n        y = 0\n        while y < img.shape[1]:\n            # Get tile\n            img_tile = img[x:x+tile_sz,y:y+tile_sz]\n\n            # Reduce if needed\n            if reduce > 1:\n                new_dim = (img_tile.shape[1]//reduce,img_tile.shape[0]//reduce)\n                img_tile = cv2.resize(img_tile, new_dim, interpolation = cv2.INTER_AREA)\n\n            # Save tile\n            save_path = \"%s_%d_%d.png\" %(img_id, x//reduce, y//reduce)\n            Image.fromarray(img_tile).save(save_path)\n            y += tile_sz\n        x += tile_sz\n\n    # Return dimension after reduction\n    final_x = ((x-tile_sz)//tile_sz)*(tile_sz//reduce) + img_tile.shape[0]\n    final_y = ((y-tile_sz)//tile_sz)*(tile_sz//reduce) + img_tile.shape[1]\n    return (final_x, final_y, 3)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Then, we must reconstruct the image from resampled tiles that we saved. For that purpose, we apply the following `_reconstruct_img` function.","metadata":{}},{"cell_type":"code","source":"def _reconstruct_img(img_id, tile_sz, shape):\n    \"\"\"\n    Reconstruct image from reduced tiles.\n    \"\"\"\n\n    img = np.zeros(shape, dtype=np.uint8)\n    if VERBOSE:\n        print(\"Reconstructed image:\", shape)\n    x = 0\n    while x < shape[0]:\n        y = 0\n        while y < shape[1]:\n            tile_path = \"%s_%d_%d.png\" %(img_id, x, y)\n            img_tile = np.asarray(Image.open(tile_path))\n            img[x:x+tile_sz,y:y+tile_sz] = img_tile\n            os.remove(tile_path) # Tiles are deleted when read\n            y += tile_sz\n        x += tile_sz\n    return img","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"All put together, the result is the following `load_resize` function.","metadata":{}},{"cell_type":"code","source":"def load_resize(idx, reduce):\n    \"\"\"\n    Memory efficient WSI loading and resampling.\n    Return resampled image and initial shape.\n    \"\"\"\n    img = load_image(os.path.join(DATA_DIR,idx+'.tiff'))\n    init_shape = img.shape\n    shape = _tile_resize_save(img, idx, (MASK_SZ*REDUCTION), reduce=REDUCTION)\n    img = _reconstruct_img(idx, (MASK_SZ*REDUCTION)//REDUCTION, shape)\n    return img, init_shape","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Making Predictions\n\nIn this section, we describe how we make predictions.\n\n### Padding for the first pass\n\nAs images are very big, we must divide them into tiles. We added zero-padding to make the dimensions of the image dividable by the tile size. Ensembling predictions with different padding dimensions allows us to avoid bad predictions due to glomeruli located on edges of the tiles. This is only necessary during the first pass as the second pass will operate on pre-selected tiles.","metadata":{}},{"cell_type":"code","source":"def _get_nored_pads_1st(initW, initH, upW, upH, xa, xb, ya, yb):\n    \"\"\"\n    Get padding to remove in final mask.\n    \"\"\"\n    px = xa/(xa+xb)\n    py = ya/(ya+yb)\n    padx = upW - initW\n    pady = upH - initH\n    assert padx > 0\n    assert pady > 0\n    xa = int(px*padx)\n    xb = padx - xa\n    ya = int(py*pady)\n    yb = pady - ya\n    return xa, xb, ya, yb\n\ndef _add_padding_1st(img, p0, p1):\n    \"\"\"\n    Add padding to make the image dividable into tiles.\n    \"\"\"\n    start = time.time()\n    if VERBOSE:\n        print(\"  > Adding padding to make the image dividable into tiles...\")\n\n    # X overlap padding\n    pad0_ = TILE_SZ - img.shape[0]%TILE_SZ\n    x_pad = int(TILE_SZ*p0)\n    xa = (pad0_//2 + x_pad)\n    xb = pad0_+TILE_SZ-(pad0_//2 + x_pad)\n    pad0_lr = [xa, xb]\n\n    # Y overlap padding\n    pad1_ = TILE_SZ - img.shape[1]%TILE_SZ\n    y_pad = int(TILE_SZ*p1)\n    ya = (pad1_//2 + y_pad)\n    yb = pad1_+TILE_SZ-(pad1_//2 + y_pad)\n    pad1_lr = [ya, yb]\n\n    img = np.pad(img, [pad0_lr, pad1_lr, [0, 0]], constant_values=0)\n    if VERBOSE:\n        print(\"  > After padding:\", img.shape, \"Time =\", time.time() - start, \"s\")\n    return img, xa, xb, ya, yb","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### Dividing the Image into Tiles\n\nThe following ``_split_image_1st`` function allows us to divide images into tiles.","metadata":{}},{"cell_type":"code","source":"def _split_image_1st(img):\n    \"\"\"\n    Split image into tiles using the reshape+transpose trick.\n    Final shape = [nb_x*nb_y, TILE_SZ, TILE_SZ, 3].\n    \"\"\"\n    start = time.time()\n    if VERBOSE:\n        print(\"  > Splitting image into tiles...\")\n    assert not img.shape[0]%TILE_SZ # Check that width is OK\n    assert not img.shape[1]%TILE_SZ # Check that height is OK\n    img = img.reshape(img.shape[0]//TILE_SZ,\n                      TILE_SZ,\n                      img.shape[1]//TILE_SZ,\n                      TILE_SZ,\n                      3)\n    img = img.transpose(0,2,1,3,4).reshape(-1,TILE_SZ,TILE_SZ,3)\n    if VERBOSE:\n        print(\"  > Splitting done! Time =\", time.time() - start)\n    return img","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Then (for the first pass) we select tiles on which we will make predictions based on their color saturations.","metadata":{}},{"cell_type":"code","source":"def _select_tiles_1st(img):\n    \"\"\"\n    Select tiles for running the model.\n    \"\"\"\n    start = time.time()\n    if VERBOSE:\n        print(\"  > Selecting tiles...\")\n    if not os.path.exists('tmp'):\n        # Generate tmp directory if needed\n        os.makedirs('tmp')\n    idxs = []\n    for i, im in enumerate(img):\n        # Remove black or gray images based on saturation check\n        hsv = cv2.cvtColor(im, cv2.COLOR_BGR2HSV)\n        h, s, v = cv2.split(hsv)\n        if (s>S_TH).sum() <= P_TH or im.sum() <= P_TH: continue \n        cv2.imwrite(\"tmp/%d.png\" %(i,), im)\n        idxs.append(i)\n    if VERBOSE:\n        print(\"  > Tiles selected! Time =\", time.time() - start)\n    return idxs","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Eventually, we need a DataLoader to feed data to our model. This DataLoader is based on the `HuBMAPTestDataset1st` and the `HuBMAPTestDataset2nd` class below. As said before, the `HuBMAPTestDataset2nd` does not need the tile selection process we mentioned.","metadata":{}},{"cell_type":"code","source":"def img2tensor(img, dtype:np.dtype=np.float32):\n    if img.ndim==2: img = np.expand_dims(img,2)\n    img = np.transpose(img,(2,0,1))\n    return torch.from_numpy(img.astype(dtype, copy=False))\n\nclass HuBMAPTestDataset1st(Dataset):\n    def __init__(self, idxs):\n        self.fnames = idxs\n        \n    def __len__(self):\n        return len(self.fnames)\n    \n    def __getitem__(self, idx):\n        im = cv2.imread(\"tmp/%d.png\" %(self.fnames[idx],))\n        return img2tensor((im/255.0 - MEAN)/STD)\n\nclass HuBMAPTestDataset2nd(Dataset):\n    def __init__(self, centroids, img):\n        self.centroids = centroids\n        self.img = img\n        \n    def __len__(self):\n        return len(self.centroids)\n    \n    def __getitem__(self, idx):\n        x, y = self.centroids[idx]\n        xa, ya = round(x) - TILE_SZ_2nd//2, round(y) - TILE_SZ_2nd//2\n        xb, yb = xa + TILE_SZ_2nd, ya + TILE_SZ_2nd\n        xa = max(0, xa)\n        ya = max(0, ya)\n        im = np.zeros((TILE_SZ_2nd, TILE_SZ_2nd, 3))\n        tmp_im = self.img[xa:xb, ya:yb]\n        padx = TILE_SZ_2nd - tmp_im.shape[0]\n        pady = TILE_SZ_2nd - tmp_im.shape[1]\n        im[padx//2:padx//2+tmp_im.shape[0],pady//2:pady//2+tmp_im.shape[1]] = tmp_im\n        return img2tensor((im/255.0 - MEAN)/STD)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"The aforementioned DataLoader is generated using the following `_make_tiles_dataloader_1st` and `_make_tiles_dataloader_1st` functions.","metadata":{}},{"cell_type":"code","source":"def _make_tiles_dataloader_1st(idxs):\n    \"\"\"\n    Make tiles dataset.\n    \"\"\"\n    start = time.time()\n    ds = HuBMAPTestDataset1st(idxs)\n    dl = DataLoader(ds, BATCH_SIZE,\n                    num_workers=NUM_WORKERS,\n                    shuffle=False,\n                    pin_memory=True)\n    if VERBOSE:\n        print(\"  > Tiles dataset created! Time =\", time.time() - start)\n    return dl\n\ndef _make_tiles_dataloader_2nd(centroids, img):\n    \"\"\"\n    Make tiles dataset.\n    \"\"\"\n    start = time.time()\n    ds = HuBMAPTestDataset2nd(centroids, img)\n    dl = DataLoader(ds, BATCH_SIZE,\n                    num_workers=NUM_WORKERS,\n                    shuffle=False,\n                    pin_memory=True)\n    if VERBOSE:\n        print(\"  > Tiles dataset created! Time =\", time.time() - start)\n    return dl","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### Generating the Segmentation Mask\n\nThe segmentation mask is then generated. It is not yet binarized. The `_generate_masks_1st` and `_generate_masks_2nd` function outputs a zero-padded mask. Only after the second pass function `_generate_masks_2nd`, the mask is upsampled to match the original image.","metadata":{}},{"cell_type":"code","source":"def _generate_masks_1st(dl, idxs, n_tiles, init_sz):\n    \"\"\"\n    Generate masks.\n    \"\"\"\n    start = time.time()\n    if VERBOSE:\n        print(\"  > Generating masks...\")\n    mp = Model_pred_1st(MODELS_1st, dl, CUSTOM_RED)\n    mask = torch.zeros(n_tiles,\n                       init_sz,\n                       init_sz,\n                       dtype=torch.uint8)\n    for i, p in zip(idxs,iter(mp)): mask[i] = p.squeeze(-1)\n    if VERBOSE:\n        print(\"  > Masks generated! Time =\", time.time() - start)\n    return mask\n\ndef _generate_masks_2nd(dl, centroids, init_shape):\n    \"\"\"\n    Generate masks.\n    \"\"\"\n    start = time.time()\n    if VERBOSE:\n        print(\"  > Generating masks...\")\n    mp = Model_pred_2nd(MODELS_2nd, dl, CUSTOM_RED)\n    W, H = init_shape[:2]\n    alpha = REDUCTION*CUSTOM_RED\n    mask = torch.zeros(W+TILE_SZ_2nd*2*alpha,\n                       H+TILE_SZ_2nd*2*alpha,\n                       dtype=torch.uint8)\n    for c, p in zip(centroids, iter(mp)):\n        x, y = c\n        x1 = round(x)*alpha - (TILE_SZ_2nd*alpha)//2 + TILE_SZ_2nd*alpha\n        y1 = round(y)*alpha - (TILE_SZ_2nd*alpha)//2 + TILE_SZ_2nd*alpha\n        x2 = x1 + TILE_SZ_2nd*alpha\n        y2 = y1 + TILE_SZ_2nd*alpha\n        # print(x1, x2, y1, y2, p.shape)\n        mask[x1:x2,y1:y2] = torch.maximum(mask[x1:x2,y1:y2], p)\n    if VERBOSE:\n        print(\"  > Final mask generated! Time =\", time.time() - start)\n    return mask[TILE_SZ_2nd*alpha:-TILE_SZ_2nd*alpha,TILE_SZ_2nd*alpha:-TILE_SZ_2nd*alpha]","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"The `_generate_masks_1st` and `_generate_masks_2nd` functions are based on the `Model_pred_1st` and `Model_pred_2nd` iterator-like class that is used to generate predictions. During the second pass, predictions are upsampled to match the original image. The `Model_pred_2nd` class makes predictions on pre-selected tiles, and then only keeps glomeruli whose intersection with a center square of predefined dimensions is non-empty: this is to avoid problems encountered with detections close to edges, that are often wrong.","metadata":{}},{"cell_type":"code","source":"# Iterator-like wrapper that returns predicted masks\nclass Model_pred_1st:\n    def __init__(self, models, dl, red, half:bool=False):\n        self.models = models # List of models\n        self.dl = dl # Dataloader\n        self.half = half # Half precision\n        self.red = red # Reduction on reduced image\n        \n    def __iter__(self):\n        with torch.no_grad():\n            for x in iter(self.dl):\n                # Prepare input\n                x = x.to(device)\n                x = F.interpolate(x, scale_factor=1/self.red, mode='bilinear')\n                if self.half: x = x.half()\n\n                # Make predictions\n                py = 0.\n                for rot_flip in ROT_TTA_FLIPS: #[0,1]\n                    for model in self.models:\n                        xr = torch.rot90(x, rot_flip, [-2, -1])\n                        p = model(xr)\n                        p = torch.rot90(p, -rot_flip, [-2, -1])\n                        p = torch.sigmoid(p).detach()\n                        py += p\n                    for f in TTA_FLIPS:\n                        xf = torch.rot90(x, rot_flip, [-2, -1])\n                        xf = torch.flip(xf,f)\n                        for model in self.models:\n                            p = model(xf)\n                            p = torch.flip(p,f)\n                            p = torch.rot90(p, -rot_flip, [-2, -1])\n                            py += torch.sigmoid(p).detach()\n                        \n                py /= (1+len(TTA_FLIPS))*len(ROT_TTA_FLIPS)       \n                py /= len(self.models)\n\n                py = py.permute(0,2,3,1).float().cpu()\n\n                # Quantize probablities to save memory\n                py = (N_BINS*py).int()\n\n                # Output predictions\n                batch_size = len(py)\n                for i in range(batch_size):\n                    yield py[i]\n                    \n    def __len__(self):\n        return len(self.dl.dataset)\n\nclass Model_pred_2nd:\n    def __init__(self, models, dl, red, half:bool=False):\n        self.models = models # List of models\n        self.dl = dl # Dataloader\n        self.half = half # Half precision\n        self.red = red # Reduction on reduced image\n        \n    def __iter__(self):\n        with torch.no_grad():\n            for x in iter(self.dl):\n                # Prepare input\n                x = x.to(device)\n                x = F.interpolate(x, scale_factor=1/self.red, mode='bilinear')\n                if self.half: x = x.half()\n\n                # Make predictions\n                py = 0.\n                for rot_flip in ROT_TTA_FLIPS: #[0,1]\n                    for model in self.models:\n                        xr = torch.rot90(x, rot_flip, [-2, -1])\n                        p = model(xr)\n                        p = torch.rot90(p, -rot_flip, [-2, -1])\n                        p = torch.sigmoid(p).detach()\n                        py += p\n                    for f in TTA_FLIPS:\n                        xf = torch.rot90(x, rot_flip, [-2, -1])\n                        xf = torch.flip(xf,f)\n                        for model in self.models:\n                            p = model(xf)\n                            p = torch.flip(p,f)\n                            p = torch.rot90(p, -rot_flip, [-2, -1])\n                            py += torch.sigmoid(p).detach()\n                        \n                py /= (1+len(TTA_FLIPS))*len(ROT_TTA_FLIPS)       \n                py /= len(self.models)\n\n                # Upsample to initial shape\n                py = F.upsample(py, scale_factor=REDUCTION*self.red, mode=\"bilinear\")\n\n                py = (py > TH_2nd).permute(0,2,3,1).int().cpu()[:,:,:,0]\n                new_py = torch.zeros(py.shape).int()\n                c1 = (REDUCTION*TILE_SZ_2nd*self.red)//2 - (REDUCTION*CHECK_SZ*self.red)//2\n                c2 = c1 + REDUCTION*CHECK_SZ*self.red\n                for i, py_i in enumerate(py):\n                    label_py_i = skimage.measure.label(py_i.detach().numpy())\n                    check = label_py_i[c1:c2,c1:c2].max()\n                    for j in range(check+1):\n                        if j and (label_py_i[c1:c2,c1:c2] == j).max():\n                            new_py[i] += (torch.Tensor(label_py_i == j).int()*py_i).int()\n                            '''if random.random() > 0.9:\n                                plt.figure()\n                                plt.imshow(torch.Tensor(label_py_i == j).numpy())\n                                plt.figure()\n                                plt.imshow(py_i.float().cpu().detach().numpy())\n                                plt.figure()\n                                plt.imshow(new_py[i].float().cpu().detach().numpy())\n                                plt.figure()\n                                plt.imshow(x[i].permute(1, 2, 0).cpu().detach().numpy())'''\n                # py = new_py\n                # Output predictions\n                batch_size = len(py)\n                for i in range(batch_size):\n                    yield new_py[i]\n                    \n    def __len__(self):\n        return len(self.dl.dataset)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"The next step for the first pass is to remove padding.","metadata":{}},{"cell_type":"code","source":"def _reshape_depad_mask_1st(mask, init_shape, init_sz, p0, p1, xa, xb, ya, yb):\n    \"\"\"\n    Reshape tiled masks into a single mask and crop padding.\n    \"\"\"\n    start = time.time()\n    if VERBOSE:\n        print(\"  > Merge tiled masks into one mask and crop padding...\")\n    mask = mask.view(init_shape[0]//TILE_SZ,\n                     init_shape[1]//TILE_SZ,\n                     init_sz,\n                     init_sz).\\\n                permute(0,2,1,3).reshape(init_shape[0],\n                                         init_shape[1])\n    mask = mask[xa:-xb,ya:-yb]\n    if VERBOSE:\n        print(\"  > Mask created! Shape =\", mask.shape,\"Time =\", time.time() - start)\n    return mask","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### Saving the Segmentation Mask\n\nThe segmentation mask is then saved as tiles containing scores (it is not binarized yet).","metadata":{}},{"cell_type":"code","source":"def _save_mask_tiles_1st(mask, idx, p0, p1):\n    start = time.time()\n    if VERBOSE:\n        print(\"  > Saving tiles in HDD memory...\")\n    x = 0\n    while x < mask.shape[0]:\n        y = 0\n        while y < mask.shape[1]:\n            mask_tile = mask[x:x+MASK_SZ,y:y+MASK_SZ].numpy()\n            save_path = \"%s_%d_%d_%s_%s.png\" %(idx, x, y, str(p0), str(p1))\n            Image.fromarray(mask_tile).save(save_path)\n            y += MASK_SZ\n        x += MASK_SZ\n    if VERBOSE:\n        print(\"Tiles saved! Time =\", time.time() - start)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"All these functions are combined in `make_one_prediction_1st` and `make_one_prediction_2nd`, that aim at saving segmentation mask tiles based on an image and a given set of padding parameters for the first pass.","metadata":{}},{"cell_type":"code","source":"def make_one_prediction_1st(img, idx, img_shape, p0, p1):\n    \"\"\"\n    Predict a mask for one given image.\n    \"\"\"\n\n    init_sz = TILE_SZ\n\n    # Add padding to make the image dividable into tiles\n    img, xa, xb, ya, yb = _add_padding_1st(img, p0, p1)\n    img_shape_p = img.shape\n\n    # Split image into tiles using the reshape+transpose trick\n    # Final shape = [nb_x*nb_y, TILE_SZ, TILE_SZ, 3]\n    img = _split_image_1st(img)\n    n_tiles = img.shape[0]\n\n    # Select tiles for running the model\n    idxs = _select_tiles_1st(img)\n\n    # Make tiles dataset\n    dl = _make_tiles_dataloader_1st(idxs)\n\n    # Generate masks\n    mask = _generate_masks_1st(dl, idxs, n_tiles, init_sz)\n\n    # Reshape tiled masks into a single mask and crop padding\n    mask = _reshape_depad_mask_1st(mask, img_shape_p, init_sz,\n                               p0, p1, xa, xb, ya, yb)\n\n    # A little bit of cleaning...\n    gc.collect()\n    shutil.rmtree('tmp')\n\n    # Save tiles in HDD memory\n    _save_mask_tiles_1st(mask, idx, p0, p1)\n\ndef make_one_prediction_2nd(img, idx, img_shape, centroids):\n    \"\"\"\n    Predict a mask for one given image.\n    \"\"\"\n\n    init_sz = TILE_SZ_2nd\n\n    # Make tiles dataset\n    ds = HuBMAPTestDataset2nd(centroids, img)\n    dl = DataLoader(ds, BATCH_SIZE,\n                num_workers=NUM_WORKERS,\n                shuffle=False,\n                pin_memory=True)\n\n    # Generate masks\n    mask = _generate_masks_2nd(dl, centroids, img_shape)\n\n    return mask","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Segmentation mask tiles are generated, for all possible sets of padding parameters for the first pass, using the following `get_mask_tiles_1st` and `get_mask_tiles_2nd` functions.","metadata":{}},{"cell_type":"code","source":"def get_mask_tiles_1st(idx, p0_list, p1_list):\n    \"\"\"\n    Load a WSI and generate mask tiles.\n    Return initial shape of WSI and binarization threshold.\n    \"\"\"\n    img, init_shape = load_resize(idx, REDUCTION)\n    for p0 in p0_list:\n        for p1 in p1_list:\n            make_one_prediction_1st(img, idx, init_shape, p0, p1)\n    return init_shape, img\n\ndef get_mask_tiles_2nd(img, idx, init_shape, centroids):\n    \"\"\"\n    Load a WSI and generate mask tiles.\n    Return initial shape of WSI and binarization threshold.\n    \"\"\"\n    mask = make_one_prediction_2nd(img, idx, init_shape, centroids)\n    return mask","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Generating Final Predictions\n\nFinal predictions are generated stacking the outputs of the `make_predictions_1st` function and the `make_predictions_2nd` function through the `get_mask_tiles_1st` and `get_mask_tiles_2nd` functions. Eventually, the obtained mask is converted into RLE format.","metadata":{}},{"cell_type":"code","source":"def make_predictions_1st(idx):\n    \"\"\"\n    Generate bounding-box predictions for idx.\n    \"\"\"\n\n    # First generate mask tiles\n    init_shape, img = get_mask_tiles_1st(idx, X_OVERLAP, Y_OVERLAP)\n\n    # Then reconstruct mask from tiles\n    mask = torch.zeros(*img.shape[:2], dtype=torch.uint8)\n    x = 0\n    while x < img.shape[0]:\n        y = 0\n        while y < img.shape[1]:\n            mask_tile = 0.\n            for p0 in X_OVERLAP:\n                for p1 in Y_OVERLAP:\n                    tile_path = \"%s_%d_%d_%s_%s.png\" %(idx, x, y, str(p0), str(p1))\n                    mask_tile += torch.tensor(np.asarray(Image.open(tile_path), dtype=int))\n                    os.remove(tile_path)\n            NEW_TH = int(N_BINS*len(X_OVERLAP)*len(Y_OVERLAP)*TH_1st)\n            mask[x:x+MASK_SZ,y:y+MASK_SZ] = mask_tile>NEW_TH\n            y += MASK_SZ\n        x += MASK_SZ\n\n    #plt.figure()\n    #plt.imshow(mask[2750:3750,5000:6000])\n\n    # Get bounding boxes\n    label_mask = skimage.measure.label(mask)\n    bboxes = skimage.measure.regionprops(label_mask)\n    centroids = [x.centroid for x in bboxes]\n    return img, init_shape, centroids\n\ndef make_predictions_2nd(idx, img, init_shape, centroids):\n    \"\"\"\n    Generate bounding-box predictions for idx.\n    \"\"\"\n\n    mask = get_mask_tiles_2nd(img, idx, init_shape, centroids)\n    #plt.figure()\n    #plt.imshow(mask[11000:15000,20000:24000].float().numpy())\n    #plt.figure()\n    #plt.imshow(img[2750:3750,5000:6000])\n    # Eventually convert to rle\n    # https://www.kaggle.com/bguberfain/memory-aware-rle-encoding\n    if VERBOSE:\n        print(\"  > Converting to RLE...\")\n    rle = rle_encode_less_memory(mask.numpy())\n    del mask\n    return rle","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"The RLE conversion is done using the following `rle_encode_less_memory` function, obtained from https://www.kaggle.com/bguberfain/memory-aware-rle-encoding.\n","metadata":{}},{"cell_type":"code","source":"#https://www.kaggle.com/bguberfain/memory-aware-rle-encoding\ndef rle_encode_less_memory(img):\n    #watch out for the bug\n    pixels = img.T.flatten()\n    # This simplified method requires first and last pixel to be zero\n    pixels[0] = 0\n    pixels[-1] = 0\n    runs = np.where(pixels[1:] != pixels[:-1])[0] + 2\n    runs[1::2] -= runs[::2]\n    return ' '.join(str(x) for x in runs)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Submission\n\nHere is the final part of this kernel, dealing with the actual submission of a CSV file containing generated masks.\n\nImage names from public and private test sets are retrieved from the sample_submission.csv file that is provided. If `PUBLIC_ONLY = True`, then we only make the prediction for public test data.","metadata":{}},{"cell_type":"code","source":"df_sample = pd.read_csv('../input/hubmap-kidney-segmentation/sample_submission.csv')\nnames,preds = [],[]\nif PUBLIC_ONLY:\n    samples = ['d488c759a', 'aa05346ff','57512b7f1','3589adb90','2ec3f1bb9']\n    samples_n = [id for id in df_sample.id if id not in samples]\n\n    for x in samples_n:\n        names += [x]\n    preds += [np.NaN]*len(samples_n)\n    df_sample = df_sample.loc[df_sample.id.isin(samples)]","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Then we generate masks for each image in the test set.","metadata":{}},{"cell_type":"code","source":"for idx,row in tqdm(df_sample.iterrows(),total=len(df_sample)):\n    idx = row['id']\n    print(\"Computing predictions for image\", idx)\n    img, init_shape, centroids = make_predictions_1st(idx)\n    rle = make_predictions_2nd(idx, img, init_shape, centroids)\n    names.append(idx)\n    preds.append(rle)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"And finally we generate a submission.csv file containing all our predictions.","metadata":{}},{"cell_type":"code","source":"df = pd.DataFrame({'id': names, 'predicted': preds})\ndf.to_csv('submission.csv',index=False)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Insights\n\nThanks to our predictions, we could compute some metrics about glomeruli. In particular we were interested in the median area per patient, the median perimeter per patient and the median ratio area/(perimeter²) per patient. Here are the values we obtained based on the GT for images in the training set and based on our predictions for images in the test set.\n\n| id        | median area (µm²) | median perimeter (µm) | median area/perimeter² |\n|-----------|-------------------|-----------------------|------------------------|\n| 0486052bb | 5610              | 309                   | 0.059                  |\n| 2f6ecfcdf | 34158             | 708                   | 0.068                  |\n| 8242609fa | 32320             | 683                   | 0.069                  |\n| aaa6a05cc | 13466             | 436                   | 0.071                  |\n| b2dc8411c | 24430             | 594                   | 0.069                  |\n| b9a3865fc | 15859             | 483                   | 0.068                  |\n| cb2d976f4 | 13378             | 435                   | 0.071                  |\n| 095bf7a1f | 29992             | 651                   | 0.071                  |\n| 1e2425f28 | 8768              | 356                   | 0.069                  |\n| 26dc41664 | 28316             | 649                   | 0.067                  |\n| 4ef6695ce | 22636             | 574                   | 0.069                  |\n| 54f2eec69 | 32392             | 684                   | 0.069                  |\n| afa5e8098 | 21632             | 577                   | 0.065                  |\n| c68fe75ea | 34348             | 720                   | 0.067                  |\n| e79de561c | 33504             | 704                   | 0.068                  |\n| 2ec3f1bb9 | 19719             | 529                   | 0.070                  |\n| 3589adb90 | 28081             | 630                   | 0.071                  |\n| 57512b7f1 | 38216             | 741                   | 0.070                  |\n| aa05346ff | 55608             | 1122                  | 0.044                  |\n| d488c759a | 11376             | 403                   | 0.070                  |\n\nWe found that the median area of glomeruli was negatively correlated to the weight (correlation coefficient of -0.40), the height (-0.39) and the BMI (-0.12). The median perimeter was similarly correlated (respectively -0.46, -0.40 and -0.20). The median area/(perimeter²), describing how complex the shape of the glomeruli were, has been found to be correlated to the weight (0.22), the height (0.16) and the BMI (0.19).\n\nHowever these findings must be confirmed on more patients, as they are based on a very small sample of people that is not necessarily representative of the general population.","metadata":{}},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}