{"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":"

Sartorius

\n

Cell Instance Segmentation Challenge

\n
CREATED BY: DARIEN SCHETTLER
\n\n
\n\n---\n\n
\n\n
\n πŸ›‘   WARNING:

THIS IS A WORK IN PROGRESS
\n
\n\n\n","metadata":{}},{"cell_type":"markdown","source":"

\n\n

\n\n

TABLE OF CONTENTS

\n\n---\n\n

0    IMPORTS

\n\n---\n\n

1    BACKGROUND INFORMATION

\n\n---\n\n

2    SETUP

\n\n---\n\n

3    HELPER FUNCTIONS

\n\n---\n\n

4    DATASET CREATION AND EXPLORATION

\n\n---\n\n

5    MODELLING

\n\n---","metadata":{}},{"cell_type":"markdown","source":"
\n\n\n\n

0  IMPORTS    ⤒

","metadata":{}},{"cell_type":"code","source":"print(\"\\n... IMPORTS STARTING ...\\n\")\nprint(\"\\n\\tVERSION INFORMATION\")\n# Machine Learning and Data Science Imports\nimport tensorflow as tf; print(f\"\\t\\t– TENSORFLOW VERSION: {tf.__version__}\");\nimport tensorflow_addons as tfa; print(f\"\\t\\t– TENSORFLOW ADDONS VERSION: {tfa.__version__}\");\nimport pandas as pd; pd.options.mode.chained_assignment = None;\nimport numpy as np; print(f\"\\t\\t– NUMPY VERSION: {np.__version__}\");\nimport sklearn; print(f\"\\t\\t– SKLEARN VERSION: {sklearn.__version__}\");\nfrom sklearn.preprocessing import RobustScaler, PolynomialFeatures\nfrom sklearn.model_selection import GroupKFold;\n\n!pip install pandarallel\nfrom pandarallel import pandarallel; pandarallel.initialize();\n\n# Built In Imports\nfrom kaggle_datasets import KaggleDatasets\nfrom collections import Counter\nfrom datetime import datetime\nfrom glob import glob\nimport warnings\nimport requests\nimport imageio\nimport IPython\nimport sklearn\nimport urllib\nimport zipfile\nimport pickle\nimport random\nimport shutil\nimport string\nimport math\nimport time\nimport gzip\nimport ast\nimport sys\nimport io\nimport os\nimport gc\nimport re\n\n# Visualization Imports\nfrom matplotlib.colors import ListedColormap\nimport matplotlib.patches as patches\nimport plotly.graph_objects as go\nimport matplotlib.pyplot as plt\nfrom tqdm.notebook import tqdm; tqdm.pandas();\nimport plotly.express as px\nimport seaborn as sns\nfrom PIL import Image\nimport matplotlib; print(f\"\\t\\t– MATPLOTLIB VERSION: {matplotlib.__version__}\");\nimport plotly\nimport PIL\nimport cv2\n\n\ndef seed_it_all(seed=7):\n \"\"\" Attempt to be Reproducible \"\"\"\n os.environ['PYTHONHASHSEED'] = str(seed)\n random.seed(seed)\n np.random.seed(seed)\n tf.random.set_seed(seed)\n\n \nprint(\"\\n\\n... IMPORTS COMPLETE ...\\n\")\n \nprint(\"\\n... SEEDING FOR DETERMINISTIC BEHAVIOUR ...\\n\")\nseed_it_all()","metadata":{"execution":{"iopub.status.busy":"2021-10-14T22:36:09.749483Z","iopub.execute_input":"2021-10-14T22:36:09.749796Z","iopub.status.idle":"2021-10-14T22:36:20.150707Z","shell.execute_reply.started":"2021-10-14T22:36:09.74972Z","shell.execute_reply":"2021-10-14T22:36:20.149026Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"
\n\n\n\n

1  BACKGROUND INFORMATION    ⤒

\n\n---\n","metadata":{}},{"cell_type":"markdown","source":"

1.1 BASIC COMPETITION INFORMATION

\n\n---\n\n
PRIMARY TASK DESCRIPTION\n\n**This is a placeholder copy/paste from the OVERVIEW>DESCRIPTION page of the Kaggle competition**\n\n\nNeurological disorders, including neurodegenerative diseases such as Alzheimer's and brain tumors, are a leading cause of death and disability across the globe. However, it is hard to **quantify how well these deadly disorders respond to treatment**. One accepted method is to review neuronal cells via **light microscopy**, which is both accessible and non-invasive. Unfortunately, **segmenting individual neuronal cells in microscopic images** can be challenging and time-intensive. **Accurate instance segmentation of these cellsβ€”with the help of computer visionβ€”could lead to new and effective drug discoveries to treat the millions of people with these disorders**.\n\nCurrent solutions have **limited accuracy for neuronal cells in particular**. In internal studies to develop cell instance segmentation models, **the neuroblastoma cell line SH-SY5Y consistently exhibits the lowest precision scores out of eight different cancer cell types tested**. This could be because **neuronal cells have a very unique, irregular and concave morphology** associated with them, making them challenging to segment with commonly used mask heads.\n\nSartorius is a partner of the life science research and the biopharmaceutical industry. They empower scientists and engineers to simplify and accelerate progress in life science and bioprocessing, enabling the development of new and better therapies and more affordable medicine. They're a magnet and dynamic platform for pioneers and leading experts in the field. They bring creative minds together for a common goal: technological breakthroughs that lead to better health for more people.\n\n
\n\n---\n\n**In this competition, you’ll detect and delineate distinct objects of interest in biological images depicting neuronal cell types commonly used in the study of neurological disorders. More specifically, you'll use phase contrast microscopy images to train and test your model for instance segmentation of neuronal cells. Successful models will do this with a high level of accuracy.**\n\n---\n\n
\n\nIf successful, you'll help further research in neurobiology thanks to the collection of robust quantitative data. Researchers may be able to use this to more easily measure the effects of disease and treatment conditions on neuronal cells. As a result, new drugs could be discovered to treat the millions of people with these leading causes of death and disability.\n","metadata":{}},{"cell_type":"markdown","source":"","metadata":{}},{"cell_type":"markdown","source":"

1.2 COMPETITION EVALUATION

\n\n---\n\n
GENERAL EVALUATION INFORMATION\n\n\nThis competition is evaluated on the mean average precision at different intersection over union πΌπ‘œπ‘ˆ thresholds. The IoU of a proposed set of object pixels and a set of true object pixels is calculated as:\n\n> **πΌπ‘œπ‘ˆ(𝐴,𝐡) = 𝐴∩𝐡𝐴βˆͺ𝐡**\n\nThe metric sweeps over a range of **IoU** thresholds, at each point calculating an average precision value. The threshold values range from 0.5 to 0.95 with a step size of 0.05: \n* i.e. 0.5, 0.55, 0.6, 0.65, 0.7, 0.75, 0.8, 0.85, 0.9, 0.95\n* **In other words, at a threshold of 0.5, a predicted object is considered a \"hit\" if its intersection over union with a ground truth object is greater than 0.5**\n\nAt each threshold value **𝑑**, a precision value is calculated based on the number of true positives **𝑇𝑃**, false negatives **𝐹𝑁**, and false positives **𝐹𝑃** resulting from comparing the predicted object to all ground truth objects:\n> **(𝑇𝑃(𝑑)) / (𝑇𝑃(𝑑)+𝐹𝑃(𝑑)+𝐹𝑁(𝑑))**\n* A true positive is counted when a single predicted object matches a ground truth object with an IoU above the threshold. \n* A false positive indicates a predicted object had no associated ground truth object. A false negative indicates a ground truth object had no associated predicted object. \n* The average precision of a single image is then calculated as the mean of the above precision values at each \n\nIoU threshold:\n> PLACEHOLDER (I'LL PUT THE FORMULA IN LATER)\n\n**Lastly, the score returned by the competition metric is the mean taken over the individual average precisions of each image in the test dataset.**\n\n
SUBMISSION FILE INFORMATION\n\nIn order to reduce the submission file size, our metric uses **run-length encoding (RLE) on the pixel values**. Instead of submitting an exhaustive list of indices for your segmentation, you will submit pairs of values that contain a start position and a run length. \n* E.g. '1 3' implies starting at pixel 1 and running a total of 3 pixels 1,2,3.\n\nThe competition format requires a space delimited list of pairs. For example, **`'1 3 10 5'`** implies pixels **`1,2,3,10,11,12,13,14`** are to be included in the mask. The pixels are **one-indexed**\nand numbered from top to bottom, then left to right: \n* 1 is pixel 1,1\n* 2 is pixel 2,1\n* etc.\n\nThe metric checks that the pairs are sorted, positive, and the decoded pixel values are not duplicated. It also checks that no two predicted masks for the same image are overlapping.\n\nThe file should contain a header and have the following format. Each row in your submission represents a single predicted nucleus segmentation for the given **`ImageId`**.\n\n```\nImageId,EncodedPixels \n0114f484a16c152baa2d82fdd43740880a762c93f436c8988ac461c5c9dbe7d5,1 1 \n0999dab07b11bc85fb8464fc36c947fbd8b5d6ec49817361cb780659ca805eac,1 1 \n0999dab07b11bc85fb8464fc36c947fbd8b5d6ec49817361cb780659ca805eac,2 3 8 9 \netc...\n```\n\n**Submission files may take several minutes to process due to the size.**\n\n
\n\n
IS THIS A CODE COMPETITION?\n\n**YES**\n\n
","metadata":{}},{"cell_type":"markdown","source":"

1.3 DATASET OVERVIEW

\n\n---\n\n
GENERAL INFORMATION\n\nIn this competition we are segmenting neuronal cells in images. The training annotations are provided as run length encoded masks, and the images are in PNG format. The number of images is small, but the number of annotated objects is quite high. The hidden test set is roughly 240 images.\n\n
FILES\n\n**`train.csv`** \n- IDs and masks for all training objects. None of this metadata is provided for the test set.\n\n**`id`** \n- unique identifier for object\n\n**`annotation`**\n- run length encoded pixels for the identified neuronal cell\n\n**`width`** \n- source image width\n\n**`height`** \n- source image height\n\n**`cell_type`** \n- the cell line\n\n**`plate_time`** \n- time plate was created\n\n**`sample_date`** \n- date sample was created\n\n**`sample_id`** \n- sample identifier\n\n**`elapsed_timedelta`**\n- time since first image taken of sample\n\n**`sample_submission.csv`** \n- a sample submission file in the correct format\n\n**`train`** \n- train images in PNG format\n\n**`test`** \n- test images in PNG format. Only a few test set images are available for download; the remainder can only be accessed by your notebooks when you submit.\n\n**`train_semi_supervised`** \n- **unlabeled images offered in case you want to use additional data for a semi-supervised approach.**\n\n**`LIVECell_dataset_2021`** \n- A mirror of the data from the LIVECell dataset. \n- LIVECell is the predecessor dataset to this competition. \n- **You will find extra data for the SH-SHY5Y cell line, plus several other cell lines not covered in the competition dataset that may be of interest for transfer learning.**\n","metadata":{}},{"cell_type":"markdown","source":"
\n\n\n\n

2  SETUP    ⤒

\n\n---\n","metadata":{}},{"cell_type":"markdown","source":"

2.1 ACCELERATOR DETECTION

\n\n---\n\nIn order to use **`TPU`**, we use **`TPUClusterResolver`** for the initialization which is necessary to connect to the remote cluster and initialize cloud TPUs. Let's go over two important points\n\n1. When using TPU on Kaggle, you don't need to specify arguments for **`TPUClusterResolver`**\n2. However, on **G**oogle **C**ompute **E**ngine (**GCE**), you will need to do the following:\n\n
\n\n```python\n# The name you gave to the TPU to use\nTPU_WORKER = 'my-tpu-name'\n\n# or you can also specify the grpc path directly\n# TPU_WORKER = 'grpc://xxx.xxx.xxx.xxx:8470'\n\n# The zone you chose when you created the TPU to use on GCP.\nZONE = 'us-east1-b'\n\n# The name of the GCP project where you created the TPU to use on GCP.\nPROJECT = 'my-tpu-project'\n\ntpu = tf.distribute.cluster_resolver.TPUClusterResolver(tpu=TPU_WORKER, zone=ZONE, project=PROJECT)\n```\n\n
\n πŸ›‘   WARNING:

- Although the Tensorflow documentation says it is the project name that should be provided for the argument `project`, it is actually the Project ID, that you should provide. This can be found on the GCP project dashboard page.
\n
\n\n
\n πŸ“–   REFERENCES:

\n - Guide - Use TPUs
\n - Doc - TPUClusterResolver
\n\n
","metadata":{}},{"cell_type":"code","source":"print(f\"\\n... ACCELERATOR SETUP STARTING ...\\n\")\n\n# Detect hardware, return appropriate distribution strategy\ntry:\n # TPU detection. No parameters necessary if TPU_NAME environment variable is set. On Kaggle this is always the case.\n TPU = tf.distribute.cluster_resolver.TPUClusterResolver() \nexcept ValueError:\n TPU = None\n\nif TPU:\n print(f\"\\n... RUNNING ON TPU - {TPU.master()}...\")\n tf.config.experimental_connect_to_cluster(TPU)\n tf.tpu.experimental.initialize_tpu_system(TPU)\n strategy = tf.distribute.experimental.TPUStrategy(TPU)\nelse:\n print(f\"\\n... RUNNING ON CPU/GPU ...\")\n # Yield the default distribution strategy in Tensorflow\n # --> Works on CPU and single GPU.\n strategy = tf.distribute.get_strategy() \n\n# What Is a Replica?\n# --> A single Cloud TPU device consists of FOUR chips, each of which has TWO TPU cores. \n# --> Therefore, for efficient utilization of Cloud TPU, a program should make use of each of the EIGHT (4x2) cores. \n# --> Each replica is essentially a copy of the training graph that is run on each core and \n# trains a mini-batch containing 1/8th of the overall batch size\nN_REPLICAS = strategy.num_replicas_in_sync\n \nprint(f\"... # OF REPLICAS: {N_REPLICAS} ...\\n\")\n\nprint(f\"\\n... ACCELERATOR SETUP COMPLTED ...\\n\")","metadata":{"execution":{"iopub.status.busy":"2021-10-14T22:36:20.152794Z","iopub.execute_input":"2021-10-14T22:36:20.153226Z","iopub.status.idle":"2021-10-14T22:36:20.167939Z","shell.execute_reply.started":"2021-10-14T22:36:20.153193Z","shell.execute_reply":"2021-10-14T22:36:20.166492Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"

2.2 COMPETITION DATA ACCESS

\n\n---\n\nTPUs read data must be read directly from **G**oogle **C**loud **S**torage **(GCS)**. Kaggle provides a utility library – **`KaggleDatasets`** – which has a utility function **`.get_gcs_path`** that will allow us to access the location of our input datasets within **GCS**.

\n\n
\n πŸ“Œ   TIPS:

- If you have multiple datasets attached to the notebook, you should pass the name of a specific dataset to the `get_gcs_path()` function. In our case, the name of the dataset is the name of the directory the dataset is mounted within.

\n
","metadata":{}},{"cell_type":"code","source":"print(\"\\n... DATA ACCESS SETUP STARTED ...\\n\")\n\nif TPU:\n # Google Cloud Dataset path to training and validation images\n DATA_DIR = KaggleDatasets().get_gcs_path('sartorius-cell-instance-segmentation')\n save_locally = tf.saved_model.SaveOptions(experimental_io_device='/job:localhost')\nelse:\n # Local path to training and validation images\n DATA_DIR = \"/kaggle/input/sartorius-cell-instance-segmentation\"\n save_locally = None\n \nprint(f\"\\n... DATA DIRECTORY PATH IS:\\n\\t--> {DATA_DIR}\")\n\nprint(f\"\\n... IMMEDIATE CONTENTS OF DATA DIRECTORY IS:\")\nfor file in tf.io.gfile.glob(os.path.join(DATA_DIR, \"*\")): print(f\"\\t--> {file}\")\n\n \nprint(\"\\n\\n... DATA ACCESS SETUP COMPLETED ...\\n\")","metadata":{"execution":{"iopub.status.busy":"2021-10-14T22:36:20.169056Z","iopub.execute_input":"2021-10-14T22:36:20.170389Z","iopub.status.idle":"2021-10-14T22:36:20.200022Z","shell.execute_reply.started":"2021-10-14T22:36:20.170344Z","shell.execute_reply":"2021-10-14T22:36:20.198796Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"

2.3 LEVERAGING XLA OPTIMIZATIONS

\n\n---\n\n\n**XLA** (Accelerated Linear Algebra) is a domain-specific compiler for linear algebra that can accelerate TensorFlow models with potentially no source code changes. **The results are improvements in speed and memory usage**.\n\n
\n\nWhen a TensorFlow program is run, all of the operations are executed individually by the TensorFlow executor. Each TensorFlow operation has a precompiled GPU/TPU kernel implementation that the executor dispatches to.\n\nXLA provides us with an alternative mode of running models: it compiles the TensorFlow graph into a sequence of computation kernels generated specifically for the given model. Because these kernels are unique to the model, they can exploit model-specific information for optimization.

\n\n
\n πŸ›‘   WARNING:

- XLA can not currently compile functions where dimensions are not inferrable: that is, if it's not possible to infer the dimensions of all tensors without running the entire computation
\n
\n\n
\n πŸ“Œ   NOTE:

- XLA compilation is only applied to code that is compiled into a graph (in TF2 that's only a code inside tf.function).
- The jit_compile API has must-compile semantics, i.e. either the entire function is compiled with XLA, or an errors.InvalidArgumentError exception is thrown)\n
\n\n
\n πŸ“–   REFERENCE:

- XLA: Optimizing Compiler for Machine Learning
\n
","metadata":{}},{"cell_type":"code","source":"print(f\"\\n... XLA OPTIMIZATIONS STARTING ...\\n\")\n\nprint(f\"\\n... CONFIGURE JIT (JUST IN TIME) COMPILATION ...\\n\")\n# enable XLA optmizations (10% speedup when using @tf.function calls)\ntf.config.optimizer.set_jit(True)\n\nprint(f\"\\n... XLA OPTIMIZATIONS COMPLETED ...\\n\")","metadata":{"execution":{"iopub.status.busy":"2021-10-14T22:36:20.201546Z","iopub.execute_input":"2021-10-14T22:36:20.201703Z","iopub.status.idle":"2021-10-14T22:36:20.208698Z","shell.execute_reply.started":"2021-10-14T22:36:20.201682Z","shell.execute_reply":"2021-10-14T22:36:20.207606Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"

2.4 BASIC DATA DEFINITIONS & INITIALIZATIONS

\n\n---\n","metadata":{}},{"cell_type":"code","source":"print(\"\\n... BASIC DATA SETUP STARTING ...\\n\\n\")\n\nprint(\"\\n... SET PATH INFORMATION ..\\n\")\nLC_DIR = os.path.join(DATA_DIR, \"LIVECell_dataset_2021\")\nLC_ANN_DIR = os.path.join(LC_DIR, \"annotations\")\nLC_IMG_DIR = os.path.join(LC_DIR, \"images\")\nTRAIN_DIR = os.path.join(DATA_DIR, \"train\")\nTEST_DIR = os.path.join(DATA_DIR, \"test\")\nSEMI_DIR = os.path.join(DATA_DIR, \"train_semi_supervised\")\n\nprint(\"\\n... TRAIN DATAFRAME ...\\n\")\n\n# FIX THE TRAIN DATAFRAME (GROUP THE RLEs TOGETHER)\nTRAIN_CSV = os.path.join(DATA_DIR, \"train.csv\")\ntrain_df = pd.read_csv(TRAIN_CSV)\ntrain_df[\"img_path\"] = train_df[\"id\"].apply(lambda x: os.path.join(TRAIN_DIR, x+\".png\")) # Capture Image Path As Well\ntmp_df = train_df.drop_duplicates(subset=[\"id\", \"img_path\"]).reset_index(drop=True)\ntmp_df[\"annotation\"] = train_df.groupby(\"id\")[\"annotation\"].agg(list).reset_index(drop=True)\ntrain_df = tmp_df.copy()\ndisplay(train_df)\n\nprint(\"\\n... SS DATAFRAME ..\\n\")\nSS_CSV = os.path.join(DATA_DIR, \"sample_submission.csv\")\nss_df = pd.read_csv(SS_CSV)\nss_df[\"img_path\"] = ss_df[\"id\"].apply(lambda x: os.path.join(TEST_DIR, x+\".png\")) # Capture Image Path As Well\n\ndisplay(ss_df)\n\nprint(\"\\n\\n... BASIC DATA SETUP FINISHING ...\\n\")","metadata":{"execution":{"iopub.status.busy":"2021-10-14T22:36:20.212167Z","iopub.execute_input":"2021-10-14T22:36:20.212452Z","iopub.status.idle":"2021-10-14T22:36:20.715508Z","shell.execute_reply.started":"2021-10-14T22:36:20.212416Z","shell.execute_reply":"2021-10-14T22:36:20.713983Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"
\n\n\n\n\n\n

\n 3  HELPER FUNCTION & CLASSES    ⤒\n

\n\n---","metadata":{}},{"cell_type":"code","source":"# ref: https://www.kaggle.com/paulorzp/run-length-encode-and-decode\ndef rle_decode(mask_rle, shape):\n '''\n mask_rle: run-length as string formated (start length)\n shape: (height,width) of array to return \n Returns numpy array, 1 - mask, 0 - background\n\n '''\n s = mask_rle.split()\n starts, lengths = [np.asarray(x, dtype=int) for x in (s[0:][::2], s[1:][::2])]\n starts -= 1\n ends = starts + lengths\n img = np.zeros(shape[0]*shape[1], dtype=np.uint8)\n for lo, hi in zip(starts, ends):\n img[lo:hi] = 1\n return img.reshape(shape) # Needed to align to RLE direction\n\n\n# ref.: https://www.kaggle.com/stainsby/fast-tested-rle\ndef rle_encode(img):\n '''\n img: numpy array, 1 - mask, 0 - background\n Returns run length as string formated\n '''\n pixels = img.flatten()\n pixels = np.concatenate([[0], pixels, [0]])\n runs = np.where(pixels[1:] != pixels[:-1])[0] + 1\n runs[1::2] -= runs[::2]\n return ' '.join(str(x) for x in runs)\n\n\ndef flatten_l_o_l(nested_list):\n \"\"\" Flatten a list of lists \"\"\"\n return [item for sublist in nested_list for item in sublist]\n","metadata":{"execution":{"iopub.status.busy":"2021-10-14T22:36:20.716598Z","iopub.execute_input":"2021-10-14T22:36:20.716834Z","iopub.status.idle":"2021-10-14T22:36:20.72951Z","shell.execute_reply.started":"2021-10-14T22:36:20.716803Z","shell.execute_reply":"2021-10-14T22:36:20.728118Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"
\n\n\n\n\n\n

\n 4  DATASET CREATION AND EXPLORATION    ⤒\n

\n\n---","metadata":{}},{"cell_type":"markdown","source":"

4.1 VISUALIZE THE DATA

\n\n---\n\nLet's create a function to take a single example (row of the dataframe) and plot the resulting information\n","metadata":{}},{"cell_type":"code","source":"def get_img_and_mask(img_path, annotation, width, height, mask_only=False):\n \"\"\" Capture the relevant image array as well as the image mask \"\"\"\n img_mask = np.zeros((height, width), dtype=np.uint8)\n for i, annot in enumerate(annotation): \n img_mask = np.where(rle_decode(annot, (height, width))!=0, i, img_mask)\n \n if mask_only:\n return img_mask\n \n img = cv2.imread(img_path)[..., ::-1]\n return img[..., 0], img_mask\n\ndef plot_img_and_mask(img, mask):\n \"\"\" Function to take an image and the corresponding mask and plot\n \n Args:\n img (np.arr): 1 channel np arr representing the image of cellular structures\n mask (np.arr): 1 channel np arr representing the instance masks (incrementing by one)\n \n Returns:\n None; Plots the two arrays and overlays them to create a merged image\n \"\"\"\n plt.figure(figsize=(20,10))\n \n plt.subplot(1,3,1)\n _img = np.tile(np.expand_dims(img, axis=-1), 3)\n plt.imshow(_img)\n plt.axis(False)\n plt.title(\"Cell Image\", fontweight=\"bold\")\n \n plt.subplot(1,3,2)\n _mask = np.zeros_like(_img)\n _mask[..., 0] = mask\n plt.imshow(mask, cmap=\"inferno\")\n plt.axis(False)\n plt.title(\"Instance Segmentation Mask\", fontweight=\"bold\")\n \n merged = cv2.addWeighted(_img, 0.75, np.clip(_mask, 0, 1)*255, 0.25, 0.0,)\n plt.subplot(1,3,3)\n plt.imshow(merged)\n plt.axis(False)\n plt.title(\"Cell Image w/ Instance Segmentation Mask Overlay\", fontweight=\"bold\")\n \n plt.tight_layout()\n plt.show()\n \n \nfor i in range(0, 10, 2):\n print(f\"\\n\\n\\n... RELEVANT DATAFRAME ROW - INDEX={i} ...\\n\")\n display(train_df.iloc[i:i+1])\n img, msk = get_img_and_mask(**train_df[[\"img_path\", \"annotation\", \"width\", \"height\"]].iloc[i].to_dict())\n plot_img_and_mask(img, msk)","metadata":{"execution":{"iopub.status.busy":"2021-10-14T22:41:39.537418Z","iopub.execute_input":"2021-10-14T22:41:39.537692Z","iopub.status.idle":"2021-10-14T22:41:42.698879Z","shell.execute_reply.started":"2021-10-14T22:41:39.537664Z","shell.execute_reply":"2021-10-14T22:41:42.69777Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def save_mask(row):\n msk = get_img_and_mask(**row[[\"img_path\", \"annotation\", \"width\", \"height\"]].to_dict(), mask_only=True)\n np.savez(f\"{row.id}\", msk)\n\ntrain_df.parallel_apply(save_mask, axis=1)","metadata":{"execution":{"iopub.status.busy":"2021-10-14T22:42:31.060124Z","iopub.execute_input":"2021-10-14T22:42:31.060367Z","iopub.status.idle":"2021-10-14T22:42:44.648335Z","shell.execute_reply.started":"2021-10-14T22:42:31.060344Z","shell.execute_reply":"2021-10-14T22:42:44.646993Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}