{"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":"# Goal:\n\n* Implement the family of Efficient-nets (b0 - b7) from scratch in PyTorch.\n* Evaluate the performance on [Cassava Leaf Disease Classification](https://www.kaggle.com/competitions/cassava-leaf-disease-classification/data?select=train.csv)\n* Compare the performance with standard PyTorch Efficient-nets.\n\n## Intution behind Efficient Net:\n\n* According to the [paper](https://arxiv.org/abs/1905.11946), The author analysed the model accuracy gain w.r.t 3 Dimensions (Depth, Width a.k.a channels, and Input Image Resolutions).\n* If you increase anyone of these 3 Dimensions, the model accuracy increases but the accuracy gain saturates after a certain point.\n* The author also observed that these 3 Dimensions are not independent of each other since if we have high resolution image then we also need more number of layers and channels to capture variety of high level of features.\n* The author proposed **compound scaling method** which uses a coefficient to uniformaly scale these 3 Dimensions.\n* The author proposed a solid baseline model in which he incorporated several tricks to have maximum accuracy gain with low number of parameters. These tricks are discussed in **Additional Information**.\n* The baseline model is called Efficient-b0 and then author uniformaly scales baseline model according to the compound scaling method and creates the whole family of Efficient nets (b0-b7).\n* Compound scaling method scales the Depth, Width, and Resolution in a Principled way as described in the below equations:\n\n![effnet.png](attachment:531db4ba-c7bd-4adc-888f-f3d262805dfa.png)\n\n## Additional Information\n\n**Depthwise Separable Convolution (DSC)**: \n* DSC is one of the tricks which makes efficient nets, efficient. \n* Lets assume a input tensor with dimensions W<sub>i</sub>, H<sub>i</sub>, C<sub>i</sub> respectively width, height, and channels.\n* In normal convolution, the computation cost of operating one k x k filter on this input tensor is **W<sub>i</sub> x H<sub>i</sub> x C<sub>i</sub> x k<sup>2</sup>** , and if we apply N such filters the cost is **N times** this number.\n* Instead if we use DSC, which is Depthwise convolution followed by Pointwise convolution, the computation cost is almost reduced by **k<sup>2</sup>** times at only a small reduction in accuracy.\n\n![DSC (1).jpg](attachment:81d49d18-d72e-481f-b60c-f2dbb8b542ce.jpg)\n\n---------------------------------------------------------------------------\n\n**Squeeze and Excitation Technique (SE)**:\n\n* SE Technique works similar to the self-attention mechanism for network channels.\n* It provides a different weight for each channel.\n* These weights emphasizes the important channels and deprecates the less relevant channels.\n\n![SE block (1).jpg](attachment:a164d29f-eeba-4a7f-8ec1-c242ae8901f5.jpg)\n\n------------------------------------------------------------------------------\n\n**Stochastic Depth**:\n\n* This is one of the tricks to accelarate the training of the model.\n* It is a training procedure to train short networks and use deep networks at test time.\n*  a subset of layers is randomly dropped and bypass them with the identity function. And a full network is used during testing/inference.\n\n![stochastic_depth.jpg](attachment:e056a849-66d9-48b1-8b56-80c85e55b9f5.jpg)","metadata":{},"attachments":{"81d49d18-d72e-481f-b60c-f2dbb8b542ce.jpg":{"image/jpeg":"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"},"a164d29f-eeba-4a7f-8ec1-c242ae8901f5.jpg":{"image/jpeg":"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"},"e056a849-66d9-48b1-8b56-80c85e55b9f5.jpg":{"image/jpeg":"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"},"531db4ba-c7bd-4adc-888f-f3d262805dfa.png":{"image/png":"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"}}},{"cell_type":"code","source":"# Useful Modules\nimport torch\nfrom torch import nn\nfrom math import ceil","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:34.756695Z","iopub.execute_input":"2022-07-23T07:16:34.757414Z","iopub.status.idle":"2022-07-23T07:16:36.666304Z","shell.execute_reply.started":"2022-07-23T07:16:34.757307Z","shell.execute_reply":"2022-07-23T07:16:36.665138Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"''' A simple Convolution, Batch Normalization, and Activation Class'''\n\nclass ConvBnAct(nn.Module):\n    \n    def __init__(self, n_in, n_out, kernel_size = 3, stride = 1, \n                 padding = 0, groups = 1, bn = True, act = True,\n                 bias = False\n                ):\n        \n        super(ConvBnAct, self).__init__()\n        \n        self.conv = nn.Conv2d(n_in, n_out, kernel_size = kernel_size,\n                              stride = stride, padding = padding,\n                              groups = groups, bias = bias\n                             )\n        self.batch_norm = nn.BatchNorm2d(n_out) if bn else nn.Identity()\n        self.activation = nn.SiLU() if act else nn.Identity()\n        \n    def forward(self, x):\n        \n        x = self.conv(x)\n        x = self.batch_norm(x)\n        x = self.activation(x)\n        \n        return x\n    \n#------------------------------------------------------------------------------\n\n''' Squeeze and Excitation Block '''\n\nclass SqueezeExcitation(nn.Module):\n    \n    def __init__(self, n_in, reduced_dim):\n        super(SqueezeExcitation, self).__init__()\n        \n        \n        self.se = nn.Sequential(\n            nn.AdaptiveAvgPool2d(1),\n            nn.Conv2d(n_in, reduced_dim, kernel_size=1),\n            nn.SiLU(),\n            nn.Conv2d(reduced_dim, n_in, kernel_size=1),\n            nn.Sigmoid()\n        )\n       \n    def forward(self, x):\n        \n        y = self.se(x)\n        \n        return x * y\n                                    \n#------------------------------------------------------------------------------\n\n''' Stochastic Depth Module'''\n\nclass StochasticDepth(nn.Module):\n    \n    def __init__(self, survival_prob = 0.8):\n        super(StochasticDepth, self).__init__()\n        \n        self.p =  survival_prob\n        \n    def forward(self, x):\n        \n        if not self.training:\n            return x\n        \n        binary_tensor = torch.rand(x.shape[0], 1, 1, 1, device=x.device) < self.p\n        \n        return torch.div(x, self.p) * binary_tensor\n        \n#-------------------------------------------------------------------------------\n\n''' Residual Bottleneck Block with Expansion Factor = N as defined in Mobilenet-V2 paper\n    with Squeeze and Excitation Block and Stochastic Depth. \n'''\n\nclass MBConvN(nn.Module):\n    \n    def __init__(self, n_in, n_out, kernel_size = 3, \n                 stride = 1, expansion_factor = 6,\n                 reduction = 4, # Squeeze and Excitation Block\n                 survival_prob = 0.8 # Stochastic Depth\n                ):\n        \n        super(MBConvN, self).__init__()\n        \n        self.skip_connection = (stride == 1 and n_in == n_out) \n        intermediate_channels = int(n_in * expansion_factor)\n        padding = (kernel_size - 1)//2\n        reduced_dim = int(n_in//reduction)\n        \n        self.expand = nn.Identity() if (expansion_factor == 1) else ConvBnAct(n_in, intermediate_channels, kernel_size = 1)\n        self.depthwise_conv = ConvBnAct(intermediate_channels, intermediate_channels,\n                                        kernel_size = kernel_size, stride = stride, \n                                        padding = padding, groups = intermediate_channels\n                                       )\n        self.se = SqueezeExcitation(intermediate_channels, reduced_dim = reduced_dim)\n        self.pointwise_conv = ConvBnAct(intermediate_channels, n_out, \n                                        kernel_size = 1, act = False\n                                       )\n        self.drop_layers = StochasticDepth(survival_prob = survival_prob)\n        \n    def forward(self, x):\n        \n        residual = x\n        \n        x = self.expand(x)\n        x = self.depthwise_conv(x)\n        x = self.se(x)\n        x = self.pointwise_conv(x)\n        \n        if self.skip_connection:\n            x = self.drop_layers(x)\n            x += residual\n        \n        return x\n    \n\n#----------------------------------------------------------------------------------------------\n\n'''Efficient-net Class'''\n\nclass EfficientNet(nn.Module):\n    \n    '''Generic Efficient net class which takes width multiplier, Depth multiplier, and Survival Prob.'''\n    \n    def __init__(self, width_mult = 1, depth_mult = 1, \n                 dropout_rate = 0.2, num_classes = 1000):\n        super(EfficientNet, self).__init__()\n        \n        last_channel = ceil(1280 * width_mult)\n        self.features = self._feature_extractor(width_mult, depth_mult, last_channel)\n        self.avgpool = nn.AdaptiveAvgPool2d(1)\n        self.classifier = nn.Sequential(\n            nn.Dropout(dropout_rate),\n            nn.Linear(last_channel, num_classes)\n        )\n        \n    def forward(self, x):\n        \n        x = self.features(x)\n        x = self.avgpool(x)\n        x = self.classifier(x.view(x.shape[0], -1))\n        \n        return x\n    \n        \n    def _feature_extractor(self, width_mult, depth_mult, last_channel):\n        \n        channels = 4*ceil(int(32*width_mult) / 4)\n        layers = [ConvBnAct(3, channels, kernel_size = 3, stride = 2, padding = 1)]\n        in_channels = channels\n        \n        kernels = [3, 3, 5, 3, 5, 5, 3]\n        expansions = [1, 6, 6, 6, 6, 6, 6]\n        num_channels = [16, 24, 40, 80, 112, 192, 320]\n        num_layers = [1, 2, 2, 3, 3, 4, 1]\n        strides =[1, 2, 2, 2, 1, 2, 1]\n        \n        # Scale channels and num_layers according to width and depth multipliers.\n        scaled_num_channels = [4*ceil(int(c*width_mult) / 4) for c in num_channels]\n        scaled_num_layers = [int(d * depth_mult) for d in num_layers]\n\n        \n        for i in range(len(scaled_num_channels)):\n             \n            layers += [MBConvN(in_channels if repeat==0 else scaled_num_channels[i], \n                               scaled_num_channels[i],\n                               kernel_size = kernels[i],\n                               stride = strides[i] if repeat==0 else 1, \n                               expansion_factor = expansions[i]\n                              )\n                       for repeat in range(scaled_num_layers[i])\n                      ]\n            in_channels = scaled_num_channels[i]\n        \n        layers.append(ConvBnAct(in_channels, last_channel, kernel_size = 1, stride = 1, padding = 0))\n    \n        return nn.Sequential(*layers)","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:36.668987Z","iopub.execute_input":"2022-07-23T07:16:36.669724Z","iopub.status.idle":"2022-07-23T07:16:36.712887Z","shell.execute_reply.started":"2022-07-23T07:16:36.669676Z","shell.execute_reply":"2022-07-23T07:16:36.711519Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Compound scaling factors for efficient-net family.\nefficient_net_config = {\n    # tuple of width multiplier, depth multiplier, resolution, and Survival Prob\n    \"b0\" : (1.0, 1.0, 224, 0.2),\n    \"b1\" : (1.0, 1.1, 240, 0.2),\n    \"b2\" : (1.1, 1.2, 260, 0.3),\n    \"b3\" : (1.2, 1.4, 300, 0.3),\n    \"b4\" : (1.4, 1.8, 380, 0.4),\n    \"b5\" : (1.6, 2.2, 456, 0.4),\n    \"b6\" : (1.8, 2.6, 528, 0.5),\n    \"b7\" : (2.0, 3.1, 600, 0.5)\n}","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:36.714748Z","iopub.execute_input":"2022-07-23T07:16:36.715270Z","iopub.status.idle":"2022-07-23T07:16:36.724381Z","shell.execute_reply.started":"2022-07-23T07:16:36.715216Z","shell.execute_reply":"2022-07-23T07:16:36.723018Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def test():\n    version = 'b0'\n    width_mult, depth_mult, res, dropout_rate = efficient_net_config[version]\n    net = EfficientNet(width_mult, depth_mult, dropout_rate)\n    x = torch.rand(1, 3, res, res)\n    y = net(x)\n    print(y.size())\n    \ntest()","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:37.190316Z","iopub.execute_input":"2022-07-23T07:16:37.190711Z","iopub.status.idle":"2022-07-23T07:16:37.560062Z","shell.execute_reply.started":"2022-07-23T07:16:37.190678Z","shell.execute_reply":"2022-07-23T07:16:37.558829Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# [Cassava Leaf Disease Classification](https://www.kaggle.com/competitions/cassava-leaf-disease-classification/data?select=train.csv)","metadata":{}},{"cell_type":"code","source":"# Useful Imports\nimport os, random\nimport cv2\nimport pandas as pd\nimport numpy as np\nfrom matplotlib import pyplot as plt\n\nimport json","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:39.304526Z","iopub.execute_input":"2022-07-23T07:16:39.308017Z","iopub.status.idle":"2022-07-23T07:16:39.491405Z","shell.execute_reply.started":"2022-07-23T07:16:39.307966Z","shell.execute_reply":"2022-07-23T07:16:39.490147Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"DATA_DIR = '../input/cassava-leaf-disease-classification'","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:40.154704Z","iopub.execute_input":"2022-07-23T07:16:40.155863Z","iopub.status.idle":"2022-07-23T07:16:40.161275Z","shell.execute_reply.started":"2022-07-23T07:16:40.155812Z","shell.execute_reply":"2022-07-23T07:16:40.160276Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Reading the labels.\nf = open(os.path.join(DATA_DIR, 'label_num_to_disease_map.json'))\n  \n# returns JSON object as \n# a dictionary\nlabels = json.load(f)\n  \n# Iterating through the json\n# list\nlabels","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:41.002735Z","iopub.execute_input":"2022-07-23T07:16:41.003912Z","iopub.status.idle":"2022-07-23T07:16:41.022615Z","shell.execute_reply.started":"2022-07-23T07:16:41.003861Z","shell.execute_reply":"2022-07-23T07:16:41.021605Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df = pd.read_csv(os.path.join(DATA_DIR, 'train.csv'))\n\ndf['disease'] = df['label'].apply(lambda x : labels[str(x)])\ndf.tail()","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:41.545590Z","iopub.execute_input":"2022-07-23T07:16:41.546681Z","iopub.status.idle":"2022-07-23T07:16:41.602299Z","shell.execute_reply.started":"2022-07-23T07:16:41.546632Z","shell.execute_reply":"2022-07-23T07:16:41.601246Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import seaborn as sns\n\nplt.xticks(rotation=25)\nsns.countplot(df['disease'])","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:42.675475Z","iopub.execute_input":"2022-07-23T07:16:42.676149Z","iopub.status.idle":"2022-07-23T07:16:43.334631Z","shell.execute_reply.started":"2022-07-23T07:16:42.676103Z","shell.execute_reply":"2022-07-23T07:16:43.333587Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Helper functions","metadata":{}},{"cell_type":"code","source":"from torch.utils.data import Dataset\nfrom PIL import Image\n\n# Dataset Class\nclass CassavaLeafDataset(Dataset):\n    \n    def __init__(self, data_path, metadata_csv, transform = None):\n        super(CassavaLeafDataset, self).__init__()\n        \n        self.df = metadata_csv #pd.read_csv(metadata_csv_path)\n        self.data_path = data_path\n        self.transform = transform\n        \n    def __len__(self):\n        return len(self.df)\n    \n    def __getitem__(self, index):\n        img_name = self.df['image_id'].iloc[index]\n        class_label = torch.tensor(int(self.df['label'].iloc[index]))\n        \n        img = Image.open(os.path.join(self.data_path, img_name))\n        \n        if self.transform is not None:\n            img = self.transform(img)\n            \n        return (img, class_label)\n    \n#------------------------------------------------------------------------------------------\n\nimport torchvision\n\ndef imshow(inp, title=None):\n    \"\"\"Imshow for Tensor.\"\"\"\n    inp = inp.numpy().transpose((1, 2, 0))\n    mean = np.array([0.485, 0.456, 0.406])\n    std = np.array([0.229, 0.224, 0.225])\n    inp = std * inp + mean\n    inp = np.clip(inp, 0, 1)\n    plt.figure(figsize = (12, 12))\n    plt.imshow(inp)\n    if title is not None:\n        plt.title(title)\n    plt.pause(0.001)  # pause a bit so that plots are updated\n    \n#-----------------------------------------------------------------------------------------\n\n# Model Performance on test data\ndef calculate_loss_and_accuracy(model, dataloader, size_of_dataset, criterion):\n    \n    # Now set model to validation mode.\n    running_loss = 0\n    running_accuracy = 0\n    \n     # Processing the Test Loader\n    for (inputs, labels) in dataloader:\n        \n        # Load data to device.\n        inputs = inputs.to(device)\n        labels = labels.to(device)\n        \n        # Outputs\n        outputs = model(inputs)\n        _ , preds = torch.max(outputs, 1)\n        \n        # Outputs\n        outputs = model(inputs)\n        _ , preds = torch.max(outputs, 1)\n        \n        # Loss and Backpropagation.\n        loss = criterion(outputs, labels)\n        \n        # Statistics\n        running_loss += loss.item()*inputs.size(0)\n        running_accuracy += torch.sum(preds == labels.data)\n        \n    epoch_loss = running_loss/size_of_dataset\n    epoch_accuracy = running_accuracy/size_of_dataset\n    \n    return epoch_loss, epoch_accuracy\n\n#------------------------------------------------------------------------------------------------\nimport copy\n\ndef train(model, criterion, optimizer, scheduler, num_of_epochs):\n    \n    best_model_wts = copy.deepcopy(model.state_dict())\n    best_acc = 0.0\n    #track_training_loss = []\n    #track_val_loss = []\n\n    for epoch in range(num_of_epochs):\n\n        print(f'\\nEpoch {epoch + 1}/{num_of_epochs}')\n        print('-'*30)\n\n        model.train() # Setting model to train.\n        running_loss = 0\n        running_accuracy = 0\n\n        # Processing the Train Loader\n        for (inputs, labels) in train_loader:\n\n            # Load data to device.\n            inputs = inputs.to(device)\n            labels = labels.to(device)\n\n            optimizer.zero_grad() # zero the parameter gradients\n\n            # Outputs\n            outputs = model(inputs)\n            _ , preds = torch.max(outputs, 1)\n\n            # Loss and Backpropagation.\n            loss = criterion(outputs, labels)\n            loss.backward()\n            optimizer.step()\n\n            # Statistics\n            running_loss += loss.item()*inputs.size(0)\n            running_accuracy += torch.sum(preds == labels.data)\n        \n        scheduler.step()\n        epoch_loss = running_loss/len(train_dataset)\n        epoch_accuracy = running_accuracy/len(train_dataset)\n        #track_training_loss.append(epoch_loss) # Loss Tracking\n\n        print(f'Training Loss: {epoch_loss:.4f} Training Acc.: {epoch_accuracy:.4f}')\n\n        # Now set model to validation mode.\n        model.eval()\n\n        val_loss, val_accuracy = calculate_loss_and_accuracy(model, val_loader, len(val_dataset), criterion)\n\n        if val_accuracy > best_acc:\n            print(\"Found better model...\")\n            print('Updating the model weights....\\n')\n            print(f'Val Loss: {val_loss:.4f} Val Acc.: {val_accuracy:.4f}\\n')\n\n            best_acc = val_accuracy\n            best_model_wts = copy.deepcopy(model.state_dict())\n     \n    model.load_state_dict(best_model_wts) # update model\n    \n    return  model","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:44.720162Z","iopub.execute_input":"2022-07-23T07:16:44.720531Z","iopub.status.idle":"2022-07-23T07:16:44.969427Z","shell.execute_reply.started":"2022-07-23T07:16:44.720499Z","shell.execute_reply":"2022-07-23T07:16:44.968327Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Data Preparation","metadata":{}},{"cell_type":"code","source":"# Divide the data into training and validation (80-20 split for each category)\ntrain_df = pd.DataFrame(columns = ['image_id', 'label'])\nval_df = pd.DataFrame(columns = ['image_id', 'label'])\n\nfor label in df['label'].unique():\n    temp_df = df[df['label'] == label].reset_index(drop = True)\n    num_of_train_data = len(temp_df)*80//100\n    train_df = train_df.append(temp_df.iloc[0:num_of_train_data].reset_index(drop=True), ignore_index = True)\n    val_df = val_df.append(temp_df.iloc[num_of_train_data:].reset_index(drop=True), ignore_index = True)","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:46.134871Z","iopub.execute_input":"2022-07-23T07:16:46.135918Z","iopub.status.idle":"2022-07-23T07:16:46.175354Z","shell.execute_reply.started":"2022-07-23T07:16:46.135865Z","shell.execute_reply":"2022-07-23T07:16:46.174231Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Train and Valid Data Distribution\nfig, ax =plt.subplots(1,2, figsize=(8, 8))\nsns.countplot(train_df['label'], ax=ax[0])\nsns.countplot(val_df['label'], ax=ax[1])\n\n# set the spacing between subplots\nplt.subplots_adjust(wspace=0.4, hspace=0.4)","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:47.859285Z","iopub.execute_input":"2022-07-23T07:16:47.859717Z","iopub.status.idle":"2022-07-23T07:16:48.211590Z","shell.execute_reply.started":"2022-07-23T07:16:47.859671Z","shell.execute_reply":"2022-07-23T07:16:48.210513Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Dataset and Dataloaders.\n# Hyper-params\nBATCH_SIZE = 32\nNUM_OF_CLASSES = 5\ndevice = torch.device('cuda')\n\n# Data Prepration.\nfrom torchvision import transforms\nfrom torch.utils.data import DataLoader\n\n# Just normalization for validation\ndata_transforms = {\n    'train': transforms.Compose([\n        transforms.RandomResizedCrop(224),\n        transforms.RandomHorizontalFlip(),\n        transforms.ToTensor(),\n        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n    ]),\n    'val': transforms.Compose([\n        transforms.Resize(256),\n        transforms.CenterCrop(224),\n        transforms.ToTensor(),\n        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n    ]),\n}\n\n\n# Datasets\ntrain_dataset = CassavaLeafDataset(os.path.join(DATA_DIR, 'train_images'), train_df, data_transforms['train'])\nval_dataset = CassavaLeafDataset(os.path.join(DATA_DIR, 'train_images'), val_df, data_transforms['val'])\n\n\n# Datloaders\ntrain_loader = DataLoader(dataset = train_dataset, batch_size = BATCH_SIZE, shuffle = True)\nval_loader = DataLoader(dataset = val_dataset, batch_size = BATCH_SIZE, shuffle = False)","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:51.757267Z","iopub.execute_input":"2022-07-23T07:16:51.757672Z","iopub.status.idle":"2022-07-23T07:16:51.770000Z","shell.execute_reply.started":"2022-07-23T07:16:51.757630Z","shell.execute_reply":"2022-07-23T07:16:51.768875Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Visualization\ndata_loader = DataLoader(dataset = train_dataset, batch_size = 4, shuffle = True)\ninputs, classes = next(iter(data_loader))\n# Make a grid from batch\nout = torchvision.utils.make_grid(inputs)\n\nimshow(out, title=[labels[str(x.item())] for x in classes])","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:54.503647Z","iopub.execute_input":"2022-07-23T07:16:54.504277Z","iopub.status.idle":"2022-07-23T07:16:55.106322Z","shell.execute_reply.started":"2022-07-23T07:16:54.504239Z","shell.execute_reply":"2022-07-23T07:16:55.105187Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Standard Pytorch Efficient-net","metadata":{}},{"cell_type":"code","source":"# Pytorch Standard Resnet-18 Model.\nfrom torchvision import models\nfrom torch.optim import lr_scheduler\nimport torch.optim as optim\n\nmodel = torch.hub.load('NVIDIA/DeepLearningExamples:torchhub', 'nvidia_efficientnet_b0', pretrained=False)\n\n# Set num of classes to 5\nmodel.classifier.fc = nn.Linear(1280, NUM_OF_CLASSES, bias = True)\n\nmodel = model.to(device) # Load model to device.\n\n# Criterion.\ncriterion = nn.CrossEntropyLoss()\noptimizer = optim.SGD(model.parameters(), lr=0.001, momentum=0.9)\n\n# Decay LR by a factor of 0.1 every 7 epochs\nexp_lr_scheduler = lr_scheduler.StepLR(optimizer, step_size=7, gamma=0.1)\n\nNUM_OF_EPOCHS = 30\n\n# Training\nbest_model = train(model = model,\n                   criterion = criterion,\n                   optimizer = optimizer,\n                   scheduler = exp_lr_scheduler,\n                   num_of_epochs = NUM_OF_EPOCHS\n                  )","metadata":{"execution":{"iopub.status.busy":"2022-07-23T07:16:59.596513Z","iopub.execute_input":"2022-07-23T07:16:59.597245Z","iopub.status.idle":"2022-07-23T08:56:11.628081Z","shell.execute_reply.started":"2022-07-23T07:16:59.597208Z","shell.execute_reply":"2022-07-23T08:56:11.626477Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Our Efficient net","metadata":{}},{"cell_type":"code","source":"# Pytorch Standard Resnet-18 Model.\nfrom torchvision import models\nfrom torch.optim import lr_scheduler\nimport torch.optim as optim\n\n# Initialize Efficientnet model\nversion = 'b0'\nwidth_mult, depth_mult, res, dropout_rate = efficient_net_config[version]\nmodel = EfficientNet(width_mult, depth_mult, dropout_rate, num_classes = NUM_OF_CLASSES)\nmodel = model.to(device) # Load model to device.\n\n\n# Criterion.\ncriterion = nn.CrossEntropyLoss()\noptimizer = optim.SGD(model.parameters(), lr=0.001, momentum=0.9)\n\n# Decay LR by a factor of 0.1 every 7 epochs\nexp_lr_scheduler = lr_scheduler.StepLR(optimizer, step_size=7, gamma=0.1)\n\nNUM_OF_EPOCHS = 15\n\n# Training\nbest_model = train(model = model,\n                   criterion = criterion,\n                   optimizer = optimizer,\n                   scheduler = exp_lr_scheduler,\n                   num_of_epochs = NUM_OF_EPOCHS\n                  )","metadata":{"execution":{"iopub.status.busy":"2022-07-23T09:02:41.280208Z","iopub.execute_input":"2022-07-23T09:02:41.280575Z","iopub.status.idle":"2022-07-23T10:21:11.408306Z","shell.execute_reply.started":"2022-07-23T09:02:41.280531Z","shell.execute_reply":"2022-07-23T10:21:11.407277Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Comparison:\n\n**Pytorch Standard Effnet-b0**:\n\n* Validation Accuracy: **0.6625**\n\n\n**Our Effnet-b0**:\n\n* Validation Accuracy: **0.7415**","metadata":{}},{"cell_type":"markdown","source":"----------------------------------------------------------------------------------------------","metadata":{"execution":{"iopub.status.busy":"2022-07-23T10:22:24.597283Z","iopub.execute_input":"2022-07-23T10:22:24.597655Z","iopub.status.idle":"2022-07-23T10:22:24.602952Z","shell.execute_reply.started":"2022-07-23T10:22:24.597619Z","shell.execute_reply":"2022-07-23T10:22:24.601476Z"}}}]}