{"cells":[{"metadata":{"trusted":true},"cell_type":"code","source":"# Importing the Packages\n\nimport pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\n%matplotlib inline\n\n\nfrom pathlib import Path","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Installing Pytorch 1.6 for FastAI v2\n\n!pip install torch==1.6.0+cu101 torchvision==0.7.0+cu101 -f https://download.pytorch.org/whl/torch_stable.html","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Installing FastAI v2\n\n!pip install -Uqq fastbook\nimport fastbook\nfastbook.setup_book()\n\nfrom fastbook import *","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Checking if cuda is available to make use of GPU\n\nimport torch\nprint(torch.cuda.is_available())\nprint(torch.cuda.get_device_name(0))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Seed everything\n\ndef seed_everything(seed):\n random.seed(seed)\n os.environ['PYTHONHASHSEED'] = str(seed)\n np.random.seed(seed)\n torch.manual_seed(seed)\n torch.cuda.manual_seed_all(seed)\n torch.cuda.manual_seed(seed)\n torch.backends.cudnn.deterministic = True\n \nseed_everything(42)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"path = Path(\"../input/aptos2019-blindness-detection\")\npath.ls()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Let's have a look at the training data\n\ntrain_df = pd.read_csv(path/\"train.csv\")\nprint(train_df.shape)\ntrain_df.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# The test data does not have any labels that we are going to predict\n\ntest_df = pd.read_csv(path/\"test.csv\")\nprint(test_df.shape)\ntest_df.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_path = path/\"train_images\"\ntrain_path.ls()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Number of images in train folder is same as number of records in the csv file. We can go ahead with creating DataBunch."},{"metadata":{"trusted":true},"cell_type":"code","source":"# Let's have a look at some of the images in the folder\n\nfig = plt.figure(figsize = (10,10))\nrows = 3\ncolumns = 3\nfor i in range(1, rows*columns+1):\n img = Image.open(train_path.ls()[i])\n fig.add_subplot(rows, columns, i)\n plt.imshow(img)\n \nplt.show()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_df['diagnosis'].value_counts()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"There is some imbalance in the classes. For now, we will proceed normally.\n
We may try some balancing in the future experiments."},{"metadata":{"trusted":true},"cell_type":"code","source":"LABEL_COLS = [0, 1, 2, 3, 4]","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_dls(size, bs):\n datablock = DataBlock(\n blocks = (ImageBlock, CategoryBlock(vocab = LABEL_COLS)),\n get_x = ColReader('id_code', pref = train_path, suff = '.png'),\n get_y = ColReader('diagnosis'),\n splitter = RandomSplitter(valid_pct = 0.2, seed= 42),\n item_tfms = Resize(size)\n )\n \n return datablock.dataloaders(train_df, bs= bs)\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"dls = get_dls(224, 64)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"dls.show_batch()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"dls.valid.show_batch()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"### Modelling"},{"metadata":{},"cell_type":"markdown","source":"A simple way to think this is that Cohen’s Kappa is a quantitative measure of reliability for two raters that are rating the same thing, corrected for how often that the raters may agree by chance.
\n
\nCohen suggested the Kappa result be interpreted as follows: values ≤ 0 as indicating no agreement and 0.01–0.20 as none to slight, 0.21–0.40 as fair, 0.41– 0.60 as moderate, 0.61–0.80 as substantial, and 0.81–1.00 as almost perfect agreement.\n"},{"metadata":{"trusted":true},"cell_type":"code","source":"ARCH = densenet121\nquadratic_kappa = CohenKappa(weights = 'quadratic')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learn = cnn_learner(dls, ARCH, metrics = [accuracy, quadratic_kappa])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learn.loss_func","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learn.lr_find()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"lr = 3e-3\nlearn.fit_one_cycle(10, lr)\nlearn.recorder.plot_loss()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learn.show_results(nrows=3, figsize=(6,8))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Plot Confusion Matrix\n\ninterp = ClassificationInterpretation.from_learner(learn)\ninterp.plot_confusion_matrix(normalize = True)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Save the model\n\nlearn.save(\"densenet121-stage-1\")\nlearn = learn.load(\"densenet121-stage-1\")","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"**If you like it , please upvote :)**"}],"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":4,"nbformat_minor":4}