{"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":"code","source":"print(\"Importing Libraries\")\nimport os\nimport gc\nimport glob\nimport cv2\nimport numpy as np \nimport pandas as pd \nimport torch\nfrom torch import nn\nimport matplotlib.pyplot as plt\nfrom torch.utils.data import Dataset, DataLoader\nfrom sklearn.model_selection import train_test_split\nfrom tqdm.notebook import tqdm\nfrom torch.optim import lr_scheduler\nimport warnings\nimport torchvision.transforms as transforms\nimport sys\nsys.path.append('../input/timm-pytorch-image-models/pytorch-image-models-master')\nimport timm\nwarnings.filterwarnings(\"ignore\")\nSIZE = 512\nclass bcolors:\n    HEADER = '\\033[95m'\n    OKBLUE = '\\033[94m'\n    OKCYAN = '\\033[96m'\n    OKGREEN = '\\033[92m'\n    WARNING = '\\033[93m'\n    FAIL = '\\033[91m'\n    ENDC = '\\033[0m'\n    BOLD = '\\033[1m'\n    UNDERLINE = '\\033[4m'\n    \npatched_images_path = \"../input/mayo-clinic-patched-images\"\ncompetition_input_path = \"../input/mayo-clinic-strip-ai\"","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2022-08-30T21:06:36.71827Z","iopub.execute_input":"2022-08-30T21:06:36.718738Z","iopub.status.idle":"2022-08-30T21:06:42.273483Z","shell.execute_reply.started":"2022-08-30T21:06:36.71865Z","shell.execute_reply":"2022-08-30T21:06:42.272116Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(\"Starting\")","metadata":{"execution":{"iopub.status.busy":"2022-08-30T21:06:42.27928Z","iopub.execute_input":"2022-08-30T21:06:42.282709Z","iopub.status.idle":"2022-08-30T21:06:42.293676Z","shell.execute_reply.started":"2022-08-30T21:06:42.282664Z","shell.execute_reply":"2022-08-30T21:06:42.29239Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"paths = np.array(glob.glob(f\"../input/mayoclinic-resized-patched-images/train_cropped/*.jpg\"))\nlfunc = np.vectorize(lambda x: os.path.basename(x)[:8])\ncorresponding_image_ids = lfunc(paths)\ndf_train_source = pd.read_csv(os.path.join(competition_input_path, \"train.csv\"))\ndf_train_source.head()","metadata":{"execution":{"iopub.status.busy":"2022-08-30T21:06:42.298768Z","iopub.execute_input":"2022-08-30T21:06:42.301219Z","iopub.status.idle":"2022-08-30T21:06:48.316971Z","shell.execute_reply.started":"2022-08-30T21:06:42.301174Z","shell.execute_reply":"2022-08-30T21:06:48.316022Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_train_source[\"patched_paths\"] = df_train_source[\"image_id\"].map(lambda x: paths[x==corresponding_image_ids])\ndf_train_source.head()","metadata":{"execution":{"iopub.status.busy":"2022-08-30T21:06:48.321906Z","iopub.execute_input":"2022-08-30T21:06:48.324577Z","iopub.status.idle":"2022-08-30T21:06:52.647424Z","shell.execute_reply.started":"2022-08-30T21:06:48.324535Z","shell.execute_reply":"2022-08-30T21:06:52.646232Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"corresponging_image_labels = []\ncorresponging_image_paths = []\ncorresponding_image_ids = []\nfor i, row in tqdm(df_train_source.iterrows()):\n    N = len(row.patched_paths)\n    corresponging_image_labels.extend([str(row.label)]*N)\n    corresponding_image_ids.extend([str(row.image_id)]*N)\n    corresponging_image_paths.extend(row.patched_paths)\ndf_train_dict = {\"image_id\": corresponding_image_ids, \"label\": corresponging_image_labels, \"path\": corresponging_image_paths}\ndf_train = pd.DataFrame(df_train_dict)\nprint(len(df_train))\ndf_train.head()","metadata":{"execution":{"iopub.status.busy":"2022-08-30T21:06:54.409084Z","iopub.execute_input":"2022-08-30T21:06:54.409466Z","iopub.status.idle":"2022-08-30T21:06:54.815471Z","shell.execute_reply.started":"2022-08-30T21:06:54.409433Z","shell.execute_reply":"2022-08-30T21:06:54.813401Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"num_CE = df_train[df_train.label == \"CE\"].label.count()\nnum_LAA = df_train[df_train.label == \"LAA\"].label.count()\nprint(f\"num_CE = {num_CE}, num_LAA = {num_LAA}\")\ntransform = transforms.Compose(\n    [\n    transforms.ToTensor(),\n    ]\n)","metadata":{"execution":{"iopub.status.busy":"2022-08-30T21:06:56.400258Z","iopub.execute_input":"2022-08-30T21:06:56.400757Z","iopub.status.idle":"2022-08-30T21:06:56.501897Z","shell.execute_reply.started":"2022-08-30T21:06:56.400709Z","shell.execute_reply":"2022-08-30T21:06:56.500801Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class ImgDataset(Dataset):\n    def __init__(self, df, transform = None):\n        self.df = df \n        self.train = 'label' in df.columns\n        self.transform = transform\n    def __len__(self):\n        return len(self.df)\n    \n    def __getitem__(self, index):\n        path = self.df.iloc[index].path\n        image = cv2.imread(path)\n        if self.transform:\n            try:\n                image = self.transform(image)\n            except:\n                try:\n                    image = image.transpose(2, 0, 1)\n                except:\n                    image = np.zeros((3, 96, 96))\n        else:\n            image = image.transpose(2, 0, 1)\n        if(self.train):\n            label = {\"CE\" : 0, \"LAA\": 1}[self.df.iloc[index].label]\n        return image, label","metadata":{"execution":{"iopub.status.busy":"2022-08-30T21:06:59.387369Z","iopub.execute_input":"2022-08-30T21:06:59.387785Z","iopub.status.idle":"2022-08-30T21:06:59.398867Z","shell.execute_reply.started":"2022-08-30T21:06:59.38775Z","shell.execute_reply":"2022-08-30T21:06:59.397414Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# --------------------------------------------------------\n# Swin Transformer V2\n# Copyright (c) 2022 Microsoft\n# Licensed under The MIT License [see LICENSE for details]\n# Written by Ze Liu\n# --------------------------------------------------------\n\nimport torch\nimport torch.nn as nn\nimport torch.nn.functional as F\nimport torch.utils.checkpoint as checkpoint\nfrom timm.models.layers import DropPath, to_2tuple, trunc_normal_\nimport numpy as np\n\n\nclass Mlp(nn.Module):\n    def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.):\n        super().__init__()\n        out_features = out_features or in_features\n        hidden_features = hidden_features or in_features\n        self.fc1 = nn.Linear(in_features, hidden_features)\n        self.act = act_layer()\n        self.fc2 = nn.Linear(hidden_features, out_features)\n        self.drop = nn.Dropout(drop)\n\n    def forward(self, x):\n        x = self.fc1(x)\n        x = self.act(x)\n        x = self.drop(x)\n        x = self.fc2(x)\n        x = self.drop(x)\n        return x\n\n\ndef window_partition(x, window_size):\n    \"\"\"\n    Args:\n        x: (B, H, W, C)\n        window_size (int): window size\n\n    Returns:\n        windows: (num_windows*B, window_size, window_size, C)\n    \"\"\"\n    B, H, W, C = x.shape\n    x = x.view(B, H // window_size, window_size, W // window_size, window_size, C)\n    windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C)\n    return windows\n\n\ndef window_reverse(windows, window_size, H, W):\n    \"\"\"\n    Args:\n        windows: (num_windows*B, window_size, window_size, C)\n        window_size (int): Window size\n        H (int): Height of image\n        W (int): Width of image\n\n    Returns:\n        x: (B, H, W, C)\n    \"\"\"\n    B = int(windows.shape[0] / (H * W / window_size / window_size))\n    x = windows.view(B, H // window_size, W // window_size, window_size, window_size, -1)\n    x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1)\n    return x\n\n\nclass WindowAttention(nn.Module):\n    r\"\"\" Window based multi-head self attention (W-MSA) module with relative position bias.\n    It supports both of shifted and non-shifted window.\n\n    Args:\n        dim (int): Number of input channels.\n        window_size (tuple[int]): The height and width of the window.\n        num_heads (int): Number of attention heads.\n        qkv_bias (bool, optional):  If True, add a learnable bias to query, key, value. Default: True\n        attn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0\n        proj_drop (float, optional): Dropout ratio of output. Default: 0.0\n        pretrained_window_size (tuple[int]): The height and width of the window in pre-training.\n    \"\"\"\n\n    def __init__(self, dim, window_size, num_heads, qkv_bias=True, attn_drop=0., proj_drop=0.,\n                 pretrained_window_size=[0, 0]):\n\n        super().__init__()\n        self.dim = dim\n        self.window_size = window_size  # Wh, Ww\n        self.pretrained_window_size = pretrained_window_size\n        self.num_heads = num_heads\n\n        self.logit_scale = nn.Parameter(torch.log(10 * torch.ones((num_heads, 1, 1))), requires_grad=True)\n\n        # mlp to generate continuous relative position bias\n        self.cpb_mlp = nn.Sequential(nn.Linear(2, 512, bias=True),\n                                     nn.ReLU(inplace=True),\n                                     nn.Linear(512, num_heads, bias=False))\n\n        # get relative_coords_table\n        relative_coords_h = torch.arange(-(self.window_size[0] - 1), self.window_size[0], dtype=torch.float32)\n        relative_coords_w = torch.arange(-(self.window_size[1] - 1), self.window_size[1], dtype=torch.float32)\n        relative_coords_table = torch.stack(\n            torch.meshgrid([relative_coords_h,\n                            relative_coords_w])).permute(1, 2, 0).contiguous().unsqueeze(0)  # 1, 2*Wh-1, 2*Ww-1, 2\n        if pretrained_window_size[0] > 0:\n            relative_coords_table[:, :, :, 0] /= (pretrained_window_size[0] - 1)\n            relative_coords_table[:, :, :, 1] /= (pretrained_window_size[1] - 1)\n        else:\n            relative_coords_table[:, :, :, 0] /= (self.window_size[0] - 1)\n            relative_coords_table[:, :, :, 1] /= (self.window_size[1] - 1)\n        relative_coords_table *= 8  # normalize to -8, 8\n        relative_coords_table = torch.sign(relative_coords_table) * torch.log2(\n            torch.abs(relative_coords_table) + 1.0) / np.log2(8)\n\n        self.register_buffer(\"relative_coords_table\", relative_coords_table)\n\n        # get pair-wise relative position index for each token inside the window\n        coords_h = torch.arange(self.window_size[0])\n        coords_w = torch.arange(self.window_size[1])\n        coords = torch.stack(torch.meshgrid([coords_h, coords_w]))  # 2, Wh, Ww\n        coords_flatten = torch.flatten(coords, 1)  # 2, Wh*Ww\n        relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :]  # 2, Wh*Ww, Wh*Ww\n        relative_coords = relative_coords.permute(1, 2, 0).contiguous()  # Wh*Ww, Wh*Ww, 2\n        relative_coords[:, :, 0] += self.window_size[0] - 1  # shift to start from 0\n        relative_coords[:, :, 1] += self.window_size[1] - 1\n        relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1\n        relative_position_index = relative_coords.sum(-1)  # Wh*Ww, Wh*Ww\n        self.register_buffer(\"relative_position_index\", relative_position_index)\n\n        self.qkv = nn.Linear(dim, dim * 3, bias=False)\n        if qkv_bias:\n            self.q_bias = nn.Parameter(torch.zeros(dim))\n            self.v_bias = nn.Parameter(torch.zeros(dim))\n        else:\n            self.q_bias = None\n            self.v_bias = None\n        self.attn_drop = nn.Dropout(attn_drop)\n        self.proj = nn.Linear(dim, dim)\n        self.proj_drop = nn.Dropout(proj_drop)\n        self.softmax = nn.Softmax(dim=-1)\n\n    def forward(self, x, mask=None):\n        \"\"\"\n        Args:\n            x: input features with shape of (num_windows*B, N, C)\n            mask: (0/-inf) mask with shape of (num_windows, Wh*Ww, Wh*Ww) or None\n        \"\"\"\n        B_, N, C = x.shape\n        qkv_bias = None\n        if self.q_bias is not None:\n            qkv_bias = torch.cat((self.q_bias, torch.zeros_like(self.v_bias, requires_grad=False), self.v_bias))\n        qkv = F.linear(input=x, weight=self.qkv.weight, bias=qkv_bias)\n        qkv = qkv.reshape(B_, N, 3, self.num_heads, -1).permute(2, 0, 3, 1, 4)\n        q, k, v = qkv[0], qkv[1], qkv[2]  # make torchscript happy (cannot use tensor as tuple)\n\n        # cosine attention\n        attn = (F.normalize(q, dim=-1) @ F.normalize(k, dim=-1).transpose(-2, -1))\n        logit_scale = torch.clamp(self.logit_scale, max=torch.log(torch.tensor(1. / 0.01))).exp()\n        attn = attn * logit_scale\n\n        relative_position_bias_table = self.cpb_mlp(self.relative_coords_table).view(-1, self.num_heads)\n        relative_position_bias = relative_position_bias_table[self.relative_position_index.view(-1)].view(\n            self.window_size[0] * self.window_size[1], self.window_size[0] * self.window_size[1], -1)  # Wh*Ww,Wh*Ww,nH\n        relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous()  # nH, Wh*Ww, Wh*Ww\n        relative_position_bias = 16 * torch.sigmoid(relative_position_bias)\n        attn = attn + relative_position_bias.unsqueeze(0)\n\n        if mask is not None:\n            nW = mask.shape[0]\n            attn = attn.view(B_ // nW, nW, self.num_heads, N, N) + mask.unsqueeze(1).unsqueeze(0)\n            attn = attn.view(-1, self.num_heads, N, N)\n            attn = self.softmax(attn)\n        else:\n            attn = self.softmax(attn)\n\n        attn = self.attn_drop(attn)\n\n        x = (attn @ v).transpose(1, 2).reshape(B_, N, C)\n        x = self.proj(x)\n        x = self.proj_drop(x)\n        return x\n\n    def extra_repr(self) -> str:\n        return f'dim={self.dim}, window_size={self.window_size}, ' \\\n               f'pretrained_window_size={self.pretrained_window_size}, num_heads={self.num_heads}'\n\n    def flops(self, N):\n        # calculate flops for 1 window with token length of N\n        flops = 0\n        # qkv = self.qkv(x)\n        flops += N * self.dim * 3 * self.dim\n        # attn = (q @ k.transpose(-2, -1))\n        flops += self.num_heads * N * (self.dim // self.num_heads) * N\n        #  x = (attn @ v)\n        flops += self.num_heads * N * N * (self.dim // self.num_heads)\n        # x = self.proj(x)\n        flops += N * self.dim * self.dim\n        return flops\n\n\nclass SwinTransformerBlock(nn.Module):\n    r\"\"\" Swin Transformer Block.\n\n    Args:\n        dim (int): Number of input channels.\n        input_resolution (tuple[int]): Input resulotion.\n        num_heads (int): Number of attention heads.\n        window_size (int): Window size.\n        shift_size (int): Shift size for SW-MSA.\n        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.\n        qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True\n        drop (float, optional): Dropout rate. Default: 0.0\n        attn_drop (float, optional): Attention dropout rate. Default: 0.0\n        drop_path (float, optional): Stochastic depth rate. Default: 0.0\n        act_layer (nn.Module, optional): Activation layer. Default: nn.GELU\n        norm_layer (nn.Module, optional): Normalization layer.  Default: nn.LayerNorm\n        pretrained_window_size (int): Window size in pre-training.\n    \"\"\"\n\n    def __init__(self, dim, input_resolution, num_heads, window_size=7, shift_size=0,\n                 mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0., drop_path=0.,\n                 act_layer=nn.GELU, norm_layer=nn.LayerNorm, pretrained_window_size=0):\n        super().__init__()\n        self.dim = dim\n        self.input_resolution = input_resolution\n        self.num_heads = num_heads\n        self.window_size = window_size\n        self.shift_size = shift_size\n        self.mlp_ratio = mlp_ratio\n        if min(self.input_resolution) <= self.window_size:\n            # if window size is larger than input resolution, we don't partition windows\n            self.shift_size = 0\n            self.window_size = min(self.input_resolution)\n        assert 0 <= self.shift_size < self.window_size, \"shift_size must in 0-window_size\"\n\n        self.norm1 = norm_layer(dim)\n        self.attn = WindowAttention(\n            dim, window_size=to_2tuple(self.window_size), num_heads=num_heads,\n            qkv_bias=qkv_bias, attn_drop=attn_drop, proj_drop=drop,\n            pretrained_window_size=to_2tuple(pretrained_window_size))\n\n        self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()\n        self.norm2 = norm_layer(dim)\n        mlp_hidden_dim = int(dim * mlp_ratio)\n        self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop)\n\n        if self.shift_size > 0:\n            # calculate attention mask for SW-MSA\n            H, W = self.input_resolution\n            img_mask = torch.zeros((1, H, W, 1))  # 1 H W 1\n            h_slices = (slice(0, -self.window_size),\n                        slice(-self.window_size, -self.shift_size),\n                        slice(-self.shift_size, None))\n            w_slices = (slice(0, -self.window_size),\n                        slice(-self.window_size, -self.shift_size),\n                        slice(-self.shift_size, None))\n            cnt = 0\n            for h in h_slices:\n                for w in w_slices:\n                    img_mask[:, h, w, :] = cnt\n                    cnt += 1\n\n            mask_windows = window_partition(img_mask, self.window_size)  # nW, window_size, window_size, 1\n            mask_windows = mask_windows.view(-1, self.window_size * self.window_size)\n            attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2)\n            attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0))\n        else:\n            attn_mask = None\n\n        self.register_buffer(\"attn_mask\", attn_mask)\n\n    def forward(self, x):\n        H, W = self.input_resolution\n        B, L, C = x.shape\n        assert L == H * W, \"input feature has wrong size\"\n\n        shortcut = x\n        x = x.view(B, H, W, C)\n\n        # cyclic shift\n        if self.shift_size > 0:\n            shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2))\n        else:\n            shifted_x = x\n\n        # partition windows\n        x_windows = window_partition(shifted_x, self.window_size)  # nW*B, window_size, window_size, C\n        x_windows = x_windows.view(-1, self.window_size * self.window_size, C)  # nW*B, window_size*window_size, C\n\n        # W-MSA/SW-MSA\n        attn_windows = self.attn(x_windows, mask=self.attn_mask)  # nW*B, window_size*window_size, C\n\n        # merge windows\n        attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C)\n        shifted_x = window_reverse(attn_windows, self.window_size, H, W)  # B H' W' C\n\n        # reverse cyclic shift\n        if self.shift_size > 0:\n            x = torch.roll(shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2))\n        else:\n            x = shifted_x\n        x = x.view(B, H * W, C)\n        x = shortcut + self.drop_path(self.norm1(x))\n\n        # FFN\n        x = x + self.drop_path(self.norm2(self.mlp(x)))\n\n        return x\n\n    def extra_repr(self) -> str:\n        return f\"dim={self.dim}, input_resolution={self.input_resolution}, num_heads={self.num_heads}, \" \\\n               f\"window_size={self.window_size}, shift_size={self.shift_size}, mlp_ratio={self.mlp_ratio}\"\n\n    def flops(self):\n        flops = 0\n        H, W = self.input_resolution\n        # norm1\n        flops += self.dim * H * W\n        # W-MSA/SW-MSA\n        nW = H * W / self.window_size / self.window_size\n        flops += nW * self.attn.flops(self.window_size * self.window_size)\n        # mlp\n        flops += 2 * H * W * self.dim * self.dim * self.mlp_ratio\n        # norm2\n        flops += self.dim * H * W\n        return flops\n\n\nclass PatchMerging(nn.Module):\n    r\"\"\" Patch Merging Layer.\n\n    Args:\n        input_resolution (tuple[int]): Resolution of input feature.\n        dim (int): Number of input channels.\n        norm_layer (nn.Module, optional): Normalization layer.  Default: nn.LayerNorm\n    \"\"\"\n\n    def __init__(self, input_resolution, dim, norm_layer=nn.LayerNorm):\n        super().__init__()\n        self.input_resolution = input_resolution\n        self.dim = dim\n        self.reduction = nn.Linear(4 * dim, 2 * dim, bias=False)\n        self.norm = norm_layer(2 * dim)\n\n    def forward(self, x):\n        \"\"\"\n        x: B, H*W, C\n        \"\"\"\n        H, W = self.input_resolution\n        B, L, C = x.shape\n        assert L == H * W, \"input feature has wrong size\"\n        assert H % 2 == 0 and W % 2 == 0, f\"x size ({H}*{W}) are not even.\"\n\n        x = x.view(B, H, W, C)\n\n        x0 = x[:, 0::2, 0::2, :]  # B H/2 W/2 C\n        x1 = x[:, 1::2, 0::2, :]  # B H/2 W/2 C\n        x2 = x[:, 0::2, 1::2, :]  # B H/2 W/2 C\n        x3 = x[:, 1::2, 1::2, :]  # B H/2 W/2 C\n        x = torch.cat([x0, x1, x2, x3], -1)  # B H/2 W/2 4*C\n        x = x.view(B, -1, 4 * C)  # B H/2*W/2 4*C\n\n        x = self.reduction(x)\n        x = self.norm(x)\n\n        return x\n\n    def extra_repr(self) -> str:\n        return f\"input_resolution={self.input_resolution}, dim={self.dim}\"\n\n    def flops(self):\n        H, W = self.input_resolution\n        flops = (H // 2) * (W // 2) * 4 * self.dim * 2 * self.dim\n        flops += H * W * self.dim // 2\n        return flops\n\n\nclass BasicLayer(nn.Module):\n    \"\"\" A basic Swin Transformer layer for one stage.\n\n    Args:\n        dim (int): Number of input channels.\n        input_resolution (tuple[int]): Input resolution.\n        depth (int): Number of blocks.\n        num_heads (int): Number of attention heads.\n        window_size (int): Local window size.\n        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.\n        qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True\n        drop (float, optional): Dropout rate. Default: 0.0\n        attn_drop (float, optional): Attention dropout rate. Default: 0.0\n        drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0\n        norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm\n        downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None\n        use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False.\n        pretrained_window_size (int): Local window size in pre-training.\n    \"\"\"\n\n    def __init__(self, dim, input_resolution, depth, num_heads, window_size,\n                 mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0.,\n                 drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False,\n                 pretrained_window_size=0):\n\n        super().__init__()\n        self.dim = dim\n        self.input_resolution = input_resolution\n        self.depth = depth\n        self.use_checkpoint = use_checkpoint\n\n        # build blocks\n        self.blocks = nn.ModuleList([\n            SwinTransformerBlock(dim=dim, input_resolution=input_resolution,\n                                 num_heads=num_heads, window_size=window_size,\n                                 shift_size=0 if (i % 2 == 0) else window_size // 2,\n                                 mlp_ratio=mlp_ratio,\n                                 qkv_bias=qkv_bias,\n                                 drop=drop, attn_drop=attn_drop,\n                                 drop_path=drop_path[i] if isinstance(drop_path, list) else drop_path,\n                                 norm_layer=norm_layer,\n                                 pretrained_window_size=pretrained_window_size)\n            for i in range(depth)])\n\n        # patch merging layer\n        if downsample is not None:\n            self.downsample = downsample(input_resolution, dim=dim, norm_layer=norm_layer)\n        else:\n            self.downsample = None\n\n    def forward(self, x):\n        for blk in self.blocks:\n            if self.use_checkpoint:\n                x = checkpoint.checkpoint(blk, x)\n            else:\n                x = blk(x)\n        if self.downsample is not None:\n            x = self.downsample(x)\n        return x\n\n    def extra_repr(self) -> str:\n        return f\"dim={self.dim}, input_resolution={self.input_resolution}, depth={self.depth}\"\n\n    def flops(self):\n        flops = 0\n        for blk in self.blocks:\n            flops += blk.flops()\n        if self.downsample is not None:\n            flops += self.downsample.flops()\n        return flops\n\n    def _init_respostnorm(self):\n        for blk in self.blocks:\n            nn.init.constant_(blk.norm1.bias, 0)\n            nn.init.constant_(blk.norm1.weight, 0)\n            nn.init.constant_(blk.norm2.bias, 0)\n            nn.init.constant_(blk.norm2.weight, 0)\n\n\nclass PatchEmbed(nn.Module):\n    r\"\"\" Image to Patch Embedding\n\n    Args:\n        img_size (int): Image size.  Default: 224.\n        patch_size (int): Patch token size. Default: 4.\n        in_chans (int): Number of input image channels. Default: 3.\n        embed_dim (int): Number of linear projection output channels. Default: 96.\n        norm_layer (nn.Module, optional): Normalization layer. Default: None\n    \"\"\"\n\n    def __init__(self, img_size=224, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None):\n        super().__init__()\n        img_size = to_2tuple(img_size)\n        patch_size = to_2tuple(patch_size)\n        patches_resolution = [img_size[0] // patch_size[0], img_size[1] // patch_size[1]]\n        self.img_size = img_size\n        self.patch_size = patch_size\n        self.patches_resolution = patches_resolution\n        self.num_patches = patches_resolution[0] * patches_resolution[1]\n\n        self.in_chans = in_chans\n        self.embed_dim = embed_dim\n\n        self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size)\n        if norm_layer is not None:\n            self.norm = norm_layer(embed_dim)\n        else:\n            self.norm = None\n\n    def forward(self, x):\n        B, C, H, W = x.shape\n        # FIXME look at relaxing size constraints\n        assert H == self.img_size[0] and W == self.img_size[1], \\\n            f\"Input image size ({H}*{W}) doesn't match model ({self.img_size[0]}*{self.img_size[1]}).\"\n        x = self.proj(x).flatten(2).transpose(1, 2)  # B Ph*Pw C\n        if self.norm is not None:\n            x = self.norm(x)\n        return x\n\n    def flops(self):\n        Ho, Wo = self.patches_resolution\n        flops = Ho * Wo * self.embed_dim * self.in_chans * (self.patch_size[0] * self.patch_size[1])\n        if self.norm is not None:\n            flops += Ho * Wo * self.embed_dim\n        return flops\n\n\nclass SwinTransformerV2(nn.Module):\n    r\"\"\" Swin Transformer\n        A PyTorch impl of : `Swin Transformer: Hierarchical Vision Transformer using Shifted Windows`  -\n          https://arxiv.org/pdf/2103.14030\n\n    Args:\n        img_size (int | tuple(int)): Input image size. Default 224\n        patch_size (int | tuple(int)): Patch size. Default: 4\n        in_chans (int): Number of input image channels. Default: 3\n        num_classes (int): Number of classes for classification head. Default: 1000\n        embed_dim (int): Patch embedding dimension. Default: 96\n        depths (tuple(int)): Depth of each Swin Transformer layer.\n        num_heads (tuple(int)): Number of attention heads in different layers.\n        window_size (int): Window size. Default: 7\n        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4\n        qkv_bias (bool): If True, add a learnable bias to query, key, value. Default: True\n        drop_rate (float): Dropout rate. Default: 0\n        attn_drop_rate (float): Attention dropout rate. Default: 0\n        drop_path_rate (float): Stochastic depth rate. Default: 0.1\n        norm_layer (nn.Module): Normalization layer. Default: nn.LayerNorm.\n        ape (bool): If True, add absolute position embedding to the patch embedding. Default: False\n        patch_norm (bool): If True, add normalization after patch embedding. Default: True\n        use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False\n        pretrained_window_sizes (tuple(int)): Pretrained window sizes of each layer.\n    \"\"\"\n\n    def __init__(self, img_size=224, patch_size=4, in_chans=3, num_classes=1000,\n                 embed_dim=96, depths=[2, 2, 6, 2], num_heads=[3, 6, 12, 24],\n                 window_size=7, mlp_ratio=4., qkv_bias=True,\n                 drop_rate=0., attn_drop_rate=0., drop_path_rate=0.1,\n                 norm_layer=nn.LayerNorm, ape=False, patch_norm=True,\n                 use_checkpoint=False, pretrained_window_sizes=[0, 0, 0, 0], **kwargs):\n        super().__init__()\n\n        self.num_classes = num_classes\n        self.num_layers = len(depths)\n        self.embed_dim = embed_dim\n        self.ape = ape\n        self.patch_norm = patch_norm\n        self.num_features = int(embed_dim * 2 ** (self.num_layers - 1))\n        self.mlp_ratio = mlp_ratio\n\n        # split image into non-overlapping patches\n        self.patch_embed = PatchEmbed(\n            img_size=img_size, patch_size=patch_size, in_chans=in_chans, embed_dim=embed_dim,\n            norm_layer=norm_layer if self.patch_norm else None)\n        num_patches = self.patch_embed.num_patches\n        patches_resolution = self.patch_embed.patches_resolution\n        self.patches_resolution = patches_resolution\n\n        # absolute position embedding\n        if self.ape:\n            self.absolute_pos_embed = nn.Parameter(torch.zeros(1, num_patches, embed_dim))\n            trunc_normal_(self.absolute_pos_embed, std=.02)\n\n        self.pos_drop = nn.Dropout(p=drop_rate)\n\n        # stochastic depth\n        dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))]  # stochastic depth decay rule\n\n        # build layers\n        self.layers = nn.ModuleList()\n        for i_layer in range(self.num_layers):\n            layer = BasicLayer(dim=int(embed_dim * 2 ** i_layer),\n                               input_resolution=(patches_resolution[0] // (2 ** i_layer),\n                                                 patches_resolution[1] // (2 ** i_layer)),\n                               depth=depths[i_layer],\n                               num_heads=num_heads[i_layer],\n                               window_size=window_size,\n                               mlp_ratio=self.mlp_ratio,\n                               qkv_bias=qkv_bias,\n                               drop=drop_rate, attn_drop=attn_drop_rate,\n                               drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])],\n                               norm_layer=norm_layer,\n                               downsample=PatchMerging if (i_layer < self.num_layers - 1) else None,\n                               use_checkpoint=use_checkpoint,\n                               pretrained_window_size=pretrained_window_sizes[i_layer])\n            self.layers.append(layer)\n\n        self.norm = norm_layer(self.num_features)\n        self.avgpool = nn.AdaptiveAvgPool1d(1)\n        self.head = nn.Linear(self.num_features, num_classes) if num_classes > 0 else nn.Identity()\n\n        self.apply(self._init_weights)\n        for bly in self.layers:\n            bly._init_respostnorm()\n\n    def _init_weights(self, m):\n        if isinstance(m, nn.Linear):\n            trunc_normal_(m.weight, std=.02)\n            if isinstance(m, nn.Linear) and m.bias is not None:\n                nn.init.constant_(m.bias, 0)\n        elif isinstance(m, nn.LayerNorm):\n            nn.init.constant_(m.bias, 0)\n            nn.init.constant_(m.weight, 1.0)\n\n    @torch.jit.ignore\n    def no_weight_decay(self):\n        return {'absolute_pos_embed'}\n\n    @torch.jit.ignore\n    def no_weight_decay_keywords(self):\n        return {\"cpb_mlp\", \"logit_scale\", 'relative_position_bias_table'}\n\n    def forward_features(self, x):\n        x = self.patch_embed(x)\n        if self.ape:\n            x = x + self.absolute_pos_embed\n        x = self.pos_drop(x)\n\n        for layer in self.layers:\n            x = layer(x)\n\n        x = self.norm(x)  # B L C\n        x = self.avgpool(x.transpose(1, 2))  # B C 1\n        x = torch.flatten(x, 1)\n        return x\n\n    def forward(self, x):\n        x = self.forward_features(x)\n        x = self.head(x)\n        return x\n\n    def flops(self):\n        flops = 0\n        flops += self.patch_embed.flops()\n        for i, layer in enumerate(self.layers):\n            flops += layer.flops()\n        flops += self.num_features * self.patches_resolution[0] * self.patches_resolution[1] // (2 ** self.num_layers)\n        flops += self.num_features * self.num_classes\n        return flops","metadata":{"execution":{"iopub.status.busy":"2022-08-30T21:07:01.631723Z","iopub.execute_input":"2022-08-30T21:07:01.635535Z","iopub.status.idle":"2022-08-30T21:07:01.936878Z","shell.execute_reply.started":"2022-08-30T21:07:01.635472Z","shell.execute_reply":"2022-08-30T21:07:01.935617Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\nprint(device)","metadata":{"execution":{"iopub.status.busy":"2022-08-30T21:07:08.802963Z","iopub.execute_input":"2022-08-30T21:07:08.803694Z","iopub.status.idle":"2022-08-30T21:07:08.884634Z","shell.execute_reply.started":"2022-08-30T21:07:08.803643Z","shell.execute_reply":"2022-08-30T21:07:08.883408Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def train_model(model, dataloaders_dict, criterion, optimizer, num_epochs, scheduler=None, train_loss_list = [], val_loss_list = [], train_acc_list = [], val_acc_list = []): \n    loss_progress = []\n    for epoch in range(num_epochs):\n        model.to(device)\n        \n        for phase in ['train', 'val']:\n            if phase == 'train':\n                model.train()\n            else:\n                model.eval()\n                \n            epoch_loss = 0.0\n            epoch_acc = 0\n            tp, fp, tn, fn = 0, 0, 0, 0\n            precision = 0\n            \n            dataloader = dataloaders_dict[phase]\n            idx = 0\n            with tqdm(dataloader, unit=\"batch\") as epoch_pbar:\n                loss_printer = []\n                for item in epoch_pbar:\n                    images = item[0].to(device).float()\n                    classes = item[1].to(device).long()\n                    optimizer.zero_grad()\n\n                    with torch.set_grad_enabled(phase == 'train'):\n                        output = model(images)\n                        loss = criterion(output, classes)\n                        _, preds = torch.max(output, 1)\n\n                        if phase == 'train':\n                            loss.backward()\n                            optimizer.step()\n                            if scheduler is not None:\n                                scheduler.step_update(epoch * num_steps + idx)\n\n                        epoch_loss += loss.item() * len(output)\n                        epoch_acc += torch.sum(preds == classes.data)\n                        \n                        tp += torch.sum( (preds == classes.data) & (preds.data==1) )\n                        tn += torch.sum( (preds == classes.data) & (preds.data==0) )\n                        fp += torch.sum( (preds != classes.data) & (preds.data==1) )\n                        fn += torch.sum( (preds != classes.data) & (preds.data==0) )\n\n                        loss_printer.append(loss.item())\n                        if len(loss_printer) == 1:\n                            mean = np.mean(loss_printer)\n                            epoch_pbar.set_postfix(loss=mean)\n                            loss_progress.append(mean)\n                            loss_printer = []\n            if phase == 'train' and scheduler is not None:\n                 scheduler.step(epoch + 1)\n\n            data_size = len(dataloader.dataset)\n            epoch_loss = epoch_loss / data_size\n            epoch_acc = epoch_acc.double() / data_size\n            f1 = tp/(tp + 1/2 * (fp + fn))\n            precision = tp/(tp+fp)\n            \n            if phase == \"train\":\n                train_loss_list.append(epoch_loss)\n                train_acc_list.append(epoch_acc.cpu())\n            else:\n                val_loss_list.append(epoch_loss)\n                val_acc_list.append(epoch_acc.cpu())\n            print(f'Epoch {epoch + 1}/{num_epochs} | {phase:^5} | Loss: {epoch_loss:.4f} | Acc: {epoch_acc:.8f}')\n            print(f\"TP = {tp} | FP = {fp} | TN = {tn} | FN = {fn}\")\n            print(f'Class 1 | {phase:^5} | P: {precision:.4f} | R: {tp/(tp+fn):.8f} |  F1: {f1:.8f}')\n            print(f'Class 0 | {phase:^5} | P: {tn/(tn+fn):.4f} | R: {tn/(tn+fp):.8f} |  F1: {tn/(tn + 1/2 * (fn + fp)):.8f}')\n        \n        torch.save(model.state_dict(), f\"swin_model_2_{epoch}.pt\")\n        print(f\"saved model\")\n        \n    return train_loss_list, val_loss_list, train_acc_list, val_acc_list","metadata":{"execution":{"iopub.status.busy":"2022-08-30T21:07:10.492697Z","iopub.execute_input":"2022-08-30T21:07:10.493227Z","iopub.status.idle":"2022-08-30T21:07:10.524083Z","shell.execute_reply.started":"2022-08-30T21:07:10.49317Z","shell.execute_reply":"2022-08-30T21:07:10.522788Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model = SwinTransformerV2(img_size=96, num_classes=2, window_size=6, patch_size=4, embed_dim = 96 )\nmodel.load_state_dict(torch.load(\"../input/mayoclinic-swinv2-patched-96-models/swin_model_1_27.pt\"))\nprint(f\"N params : {sum(p.numel() for p in model.parameters())}\")","metadata":{"execution":{"iopub.status.busy":"2022-08-30T21:07:46.088376Z","iopub.execute_input":"2022-08-30T21:07:46.08878Z","iopub.status.idle":"2022-08-30T21:07:52.076526Z","shell.execute_reply.started":"2022-08-30T21:07:46.088745Z","shell.execute_reply":"2022-08-30T21:07:52.075436Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train, val = train_test_split(df_train, test_size=0.15 , stratify = df_train.label, random_state=22)\nbatch_size = 300\ntrain_loader = DataLoader(\n    ImgDataset(train, transform = transform), \n    batch_size=batch_size, \n    shuffle=True, \n    num_workers=1\n)\nval_loader = DataLoader(\n    ImgDataset(val, transform = transform), \n    batch_size=batch_size, \n    shuffle=False, \n    num_workers=1\n)\ndataloaders_dict = {\"train\": train_loader, \"val\": val_loader}\nweights = torch.zeros(2).to(device)\nweights[0] = 1 / num_CE\nweights[1] = 1 / num_LAA\ncriterion = nn.CrossEntropyLoss(weight=weights , reduction = 'mean')","metadata":{"execution":{"iopub.status.busy":"2022-08-30T21:07:57.919425Z","iopub.execute_input":"2022-08-30T21:07:57.919858Z","iopub.status.idle":"2022-08-30T21:07:58.357095Z","shell.execute_reply.started":"2022-08-30T21:07:57.919824Z","shell.execute_reply":"2022-08-30T21:07:58.355941Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from timm.scheduler.cosine_lr import CosineLRScheduler\nn_epochs = 30\nlr = 5e-4\n\n\nlinear_scaled_lr = lr * batch_size\noptimizer = torch.optim.AdamW(model.parameters(), lr=linear_scaled_lr, weight_decay=0.05, eps=1e-8, betas=(0.9, 0.999))\n\nnum_steps = n_epochs * len(train)\nwarmup_steps = 5\nlinear_scaled_warmup_lr = 5e-7 * batch_size\nlinear_scaled_min_lr = 5e-6 * batch_size \nlr_scheduler = CosineLRScheduler(\n            optimizer,\n            t_initial=num_steps,\n            lr_min=linear_scaled_min_lr,\n            warmup_lr_init=linear_scaled_warmup_lr,\n            warmup_t=warmup_steps,\n            cycle_limit=1,\n            t_in_epochs=False\n        )\n\ntrain_loss_list, val_loss_list, train_acc_list, val_acc_list = train_model(model, dataloaders_dict, criterion, optimizer, n_epochs, lr_scheduler)","metadata":{"execution":{"iopub.status.busy":"2022-08-30T21:08:01.397602Z","iopub.execute_input":"2022-08-30T21:08:01.398031Z","iopub.status.idle":"2022-08-30T21:08:27.306055Z","shell.execute_reply.started":"2022-08-30T21:08:01.397998Z","shell.execute_reply":"2022-08-30T21:08:27.301661Z"},"trusted":true},"execution_count":null,"outputs":[]}]}