","metadata":{}},{"cell_type":"code","source":"data = train_df['genus'].value_counts().head(15)\ndata = pd.DataFrame({'Genus' : data.index,\n 'values' : data.values})\nplt.figure(figsize = (20, 10))\nsns.barplot(x='values', y = 'Genus', data = data , palette='summer_r')\nplt.show()","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"def show_images(speices):\n images = train_df.loc[train_df['genus'] == speices]['image_dir'][:6]\n i = 1\n fig = plt.figure(figsize = (18, 18))\n plt.suptitle(speices, fontsize = '30')\n for image in images:\n img = cv2.imread(image)\n ax = fig.add_subplot(2, 3, i)\n ax.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))\n ax.set_axis_off()\n i += 1\n plt.show()","metadata":{"_kg_hide-output":true,"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"show_images('Carex')","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"show_images('Astragalus')","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"show_images('Penstemon')","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"show_images('Eriogonum')","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"show_images('Erigeron')","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
\n
\nModeling
","metadata":{}},{"cell_type":"code","source":"import torch\nimport torch.nn as nn\nimport torchvision\nimport torchvision.transforms as transforms\nfrom torch.utils.data import Dataset, DataLoader\nfrom PIL import Image","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"BATCH = 128\nEPOCHS = 1\n\nLR = 0.01\nIM_SIZE = 224\n\nX_Train, Y_Train = train_df['image_dir'].values, train_df['category'].values\n\nTransform = transforms.Compose(\n [transforms.ToTensor(),\n transforms.Resize((IM_SIZE, IM_SIZE)),\n transforms.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225))])","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"class GetData(Dataset):\n def __init__(self, FNames, Labels, Transform):\n self.fnames = FNames\n self.transform = Transform\n self.labels = Labels \n \n def __len__(self):\n return len(self.fnames)\n\n def __getitem__(self, index): \n x = Image.open(self.fnames[index])\n \n if \"train\" in self.fnames[index]: \n return self.transform(x), self.labels[index]\n elif \"test\" in self.fnames[index]: \n return self.transform(x), self.fnames[index]\n \ntrainset = GetData(X_Train, Y_Train, Transform)\ntrainloader = DataLoader(trainset, batch_size=BATCH, shuffle=True)\n\nN_Classes = train_df['category'].nunique()\nnext(iter(trainloader))[0].shape\n\ndevice = 'cuda:0' if torch.cuda.is_available() else 'cpu'\nmodel = torchvision.models.densenet169(pretrained=True)","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"train_df['category'].nunique()","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"print(model.classifier.in_features) \nprint(model.classifier.out_features)\n\nfor param in model.parameters():\n param.requires_grad = False\n \nn_inputs = model.classifier.in_features\nlast_layer = nn.Linear(n_inputs, N_Classes)\nmodel.classifier = last_layer\nif torch.cuda.is_available():\n model.cuda()\nprint(model.classifier.out_features) \n\ncriterion = nn.CrossEntropyLoss()\noptimizer = torch.optim.Adam(model.classifier.parameters())","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
\n\n
\n
\nTrain & Evaluate
","metadata":{}},{"cell_type":"code","source":"training_history = {'accuracy':[],'loss':[]}\nvalidation_history = {'accuracy':[],'loss':[]}\n\nfrom tqdm import tqdm\n\ndef train(trainloader, model, criterion, optimizer, scaler, device=torch.device(\"cpu\")):\n train_acc = 0.0\n train_loss = 0.0\n for images, labels in tqdm(trainloader):\n images = images.to(device)\n labels = labels.to(device)\n optimizer.zero_grad()\n with torch.cuda.amp.autocast(enabled=True):\n output = model(images)\n loss = criterion(output, labels)\n scaler.scale(loss).backward()\n scaler.step(optimizer)\n scaler.update()\n acc = ((output.argmax(dim=1) == labels).float().mean())\n train_acc += acc\n train_loss += loss\n return train_acc/len(trainloader), train_loss/len(trainloader)","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"## Normal Evaluation\ndef evaluate(testloader, model, criterion, device=torch.device(\"cpu\")):\n eval_acc = 0.0\n eval_loss = 0.0\n for images, labels in tqdm(testloader):\n images = images.to(device)\n labels = labels.to(device)\n with torch.no_grad():\n output = model(images)\n loss = criterion(output, labels)\n\n acc = ((output.argmax(dim=1) == labels).float().mean())\n eval_acc += acc\n eval_loss += loss\n \n return eval_acc/len(testloader), eval_loss/len(testloader)","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"code","source":"%%time\n#%dirsrmal Training\nscaler = torch.cuda.amp.GradScaler(enabled=True)\nfor epoch in range(EPOCHS):\n train_acc, train_loss = train(trainloader, model, criterion, optimizer, scaler, device=device)\n eval_acc, eval_loss = evaluate(val_loader, model, criterion, device=torch.device(\"cuda\"))\n print(\"\")\n print(f\"Epoch {epoch + 1} | Train Acc: {train_acc*100} | Train Loss: {train_loss}\")\n print(f\"\\t Val Acc: {eval_acc*100} | Val Loss: {eval_loss}\")\n print(\"====\"*8)","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"","metadata":{}},{"cell_type":"markdown","source":"
\n\n
\n
\nConclusion
","metadata":{}},{"cell_type":"code","source":"X_Test = test_df['test_dir'].values\ntestset = GetData(X_Test, None, Transform)\ntestloader = DataLoader(testset, batch_size=1, shuffle=False)\n\ns_ls = []\n\nwith torch.no_grad():\n model.eval()\n for image, fname in testloader: \n image = image.to(device)\n \nlogits = model(image) \nps = torch.exp(logits) \n_, top_class = ps.topk(1, dim=1)\n \nfor pred in top_class:\n s_ls.append([fname[0].split('/')[-1][:-4], pred.item()])\n \nsub = pd.DataFrame.from_records(s_ls, columns=['Id', 'Predicted'])\nsub.head()\nsub.to_csv(\"submission.csv\", index=False)","metadata":{"trusted":true},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"# Upvote!","metadata":{}}]}
\n","metadata":{}},{"cell_type":"markdown","source":"## Content\n- **Load Dataset**\n- **JSON -> DataFrame**\n- **Mapping genus**\n- **Data Visualization**\n- **Modeling**\n- **Train & Evaluate**\n- **Conclusion**\n\n