{"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":"#### 🍀thank you  all helpers \n[@Yoshi_K ](https://www.kaggle.com/yoshikuwano/eda-non-annotated-starfish)  [@CAMARO](https://www.kaggle.com/bamps53/create-annotated-video#kln-23) \n[@tensor_choko](https://www.kaggle.com/tensorchoko/tensorflow-eda-for-beginner-jp-en/notebook)","metadata":{}},{"cell_type":"markdown","source":" In this notebook I tried to explain all the things that are needed for us to kick start our model building , I kept my comments as  begginer friendly as I can , explaining all facts like a conversation , hope it will help you to understand this interesting problem <br>\n So.. let's dive in...🏊‍🏊‍\n # 🐠🐟🐡🦑🐙🦈🐬🐳🐋🦀🐚🏊‍♀️🍀☘💺🚤⚓🏝🌊🌊 🐠🐟🐡","metadata":{}},{"cell_type":"markdown","source":"<br>\n<br>\n<br>","metadata":{}},{"cell_type":"markdown","source":"# 💢 Basic info & requirements for this competition in simple language 😅\n<br>\n<br>\n\n\n![](https://505488.smushcdn.com/2242995/wp-content/uploads/site_assets/crown_of_thorns_starfish_5457578925_.jpg?lossy=1&strip=1&webp=1)\n\n\n>**Goal** : there is one species of star fish which is eating corals and destroying the great barrier reef , we have to find that COTS named star fish (above fig. ) by creating an object detection model which can work in real time <br>\n**What we have with us to analyse and create model ?**  : well , in   our data we have <br>\n.  1. one API named inside folder greatbarrier reef <br>\n.  2. train_images : these are basically 3 videos whose images are splitted into video_0 , video_1 , video_2 folders <br>\n.  3. test, train, sample submission csvs are there <br>\n.  4. example test.npy is also there : <br>\n\n>**what we have to actually do here** : for training there is a image of corals and there is marking/annotation ( a sqare bounding box ) is given in image at specific position and size , i.e. COTS fish is marked in bounding box using annotations , what we have to do is create a model which will train based on the given annotations and while testing it should create the annotations/bounding boxes on it's own to find where the actual COTS fish is, there in the given image,\nfurther more the entire system should work in real time which means, when some sea diver is taking video clip of corals deep down sea our algorithm should detect the object from that video...this is what expected by user as per my understanding.\n\n>**what we have to submit** : as usual.. we have to predict y but this time y is  annotation and submit it in submission.csv file ..  [refer code requirements](https://www.kaggle.com/c/tensorflow-great-barrier-reef/overview/code-requirements)<br>\n>**How our submission will be evaluated** : using F2 score , i.e False positivity rate is accepted , F2= 5PR/(4P+R )\n\nif someone has any doubt or any other view regarding the question understanding or if something is missing , then please comment and share your views. because it will help me to explore the data more effectively.","metadata":{}},{"cell_type":"markdown","source":"# 💥EDA for the <strong> Great barrier Reef </strong> dataset","metadata":{}},{"cell_type":"markdown","source":"#### 💠importanting libraries ","metadata":{}},{"cell_type":"code","source":"import os\nimport cv2\nimport ast\nimport json\nimport subprocess\nimport seaborn as sns\n\nimport numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\nfrom IPython.display import Video  \n","metadata":{"execution":{"iopub.status.busy":"2022-01-04T10:35:09.469887Z","iopub.execute_input":"2022-01-04T10:35:09.470240Z","iopub.status.idle":"2022-01-04T10:35:09.977875Z","shell.execute_reply.started":"2022-01-04T10:35:09.470206Z","shell.execute_reply":"2022-01-04T10:35:09.977237Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### 💠Importing Data ","metadata":{}},{"cell_type":"code","source":"main_path= '../input/tensorflow-great-barrier-reef'","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:05.577844Z","iopub.execute_input":"2022-01-03T12:50:05.578406Z","iopub.status.idle":"2022-01-03T12:50:05.58232Z","shell.execute_reply.started":"2022-01-03T12:50:05.578369Z","shell.execute_reply":"2022-01-03T12:50:05.581485Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"tdf= pd.read_csv(main_path + '/train.csv')\ntdf.head()","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:05.583826Z","iopub.execute_input":"2022-01-03T12:50:05.58452Z","iopub.status.idle":"2022-01-03T12:50:05.63913Z","shell.execute_reply.started":"2022-01-03T12:50:05.584474Z","shell.execute_reply":"2022-01-03T12:50:05.638182Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"so train.csv explain us the parameters to connect the images in the video_0,video_1 and video_2 folders, \n`video_id` -- explain us from which folder the file is , if video_id val ==0 then it belongs to video_0 and so on..\n`sequence` -- ","metadata":{}},{"cell_type":"code","source":"tdf.info()","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:05.640927Z","iopub.execute_input":"2022-01-03T12:50:05.641191Z","iopub.status.idle":"2022-01-03T12:50:05.670914Z","shell.execute_reply.started":"2022-01-03T12:50:05.641162Z","shell.execute_reply":"2022-01-03T12:50:05.670309Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"tdf.describe()","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:05.672002Z","iopub.execute_input":"2022-01-03T12:50:05.672373Z","iopub.status.idle":"2022-01-03T12:50:05.700524Z","shell.execute_reply.started":"2022-01-03T12:50:05.672341Z","shell.execute_reply":"2022-01-03T12:50:05.699752Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### 💠 Analysis for prediction Y\nlet's see what exactly we have to predict ","metadata":{}},{"cell_type":"code","source":"Tdf=pd.read_csv(main_path + '/test.csv')\nTdf.head()","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:05.702158Z","iopub.execute_input":"2022-01-03T12:50:05.702524Z","iopub.status.idle":"2022-01-03T12:50:05.718033Z","shell.execute_reply.started":"2022-01-03T12:50:05.702494Z","shell.execute_reply":"2022-01-03T12:50:05.717132Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"hmm... annotations is the key here,\nlet's see what this feature's specifications are...🤔","metadata":{}},{"cell_type":"code","source":"tdf.annotations.iloc[133:200]","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:05.719431Z","iopub.execute_input":"2022-01-03T12:50:05.719644Z","iopub.status.idle":"2022-01-03T12:50:05.728464Z","shell.execute_reply.started":"2022-01-03T12:50:05.719618Z","shell.execute_reply":"2022-01-03T12:50:05.727346Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"a=tdf.annotations.describe()\nprint(a)\ntdf.annotations.unique()","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:05.73046Z","iopub.execute_input":"2022-01-03T12:50:05.730697Z","iopub.status.idle":"2022-01-03T12:50:05.757161Z","shell.execute_reply.started":"2022-01-03T12:50:05.730669Z","shell.execute_reply":"2022-01-03T12:50:05.756299Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### ⭕annotations per video ","metadata":{}},{"cell_type":"code","source":"for video_id in tdf['video_id'].unique():\n    print(f'video_id: {video_id}')\n    print(f'no. of images without annotations:  {sum(tdf[tdf[\"video_id\"]==video_id][\"annotations\"]== \"[]\")}')\n    print(f'no. of images with annotations:  {sum(tdf[tdf[\"video_id\"]==video_id][\"annotations\"] != \"[]\")}\\n')","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:05.75898Z","iopub.execute_input":"2022-01-03T12:50:05.7592Z","iopub.status.idle":"2022-01-03T12:50:05.791795Z","shell.execute_reply.started":"2022-01-03T12:50:05.759174Z","shell.execute_reply":"2022-01-03T12:50:05.790942Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### ⭕percentage of annatotation ","metadata":{}},{"cell_type":"code","source":"a1=2143/(4565 + 2143)*100 ; a2=2099/(6133+2099)*100 ;a3=677/(7884+677)*100 ; a=(2143 +2099+677)/23501*100\nprint(f\"total % annotations in video0 : {a1}% \\ntotal % annotations in video1: {a2}% \\ntotal % annotations in video2 : {a3}% \\nToal % annotations  : {a}%\" )","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:05.793494Z","iopub.execute_input":"2022-01-03T12:50:05.793724Z","iopub.status.idle":"2022-01-03T12:50:05.799848Z","shell.execute_reply.started":"2022-01-03T12:50:05.793695Z","shell.execute_reply":"2022-01-03T12:50:05.799127Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"ok..  we got something very bad here ..\nfor every annotation we have specific co-ordinates { x ,y, width and height} but lot's of annotations are empty `only 20.93%` images are annoted and that's not at all interesting 😂😂 bcoz it's going to force us to use some advanced algorithms to create model from very less training dataset ... \n<br>\n![download.png](attachment:4c722f68-7f1c-4869-812a-12bba1433eff.png) <br>\njust kidding 😂","metadata":{},"attachments":{"4c722f68-7f1c-4869-812a-12bba1433eff.png":{"image/png":"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"}}},{"cell_type":"markdown","source":"ok.. let's see what we have in our sample submission example","metadata":{}},{"cell_type":"markdown","source":"### 💠 Expected output annotation\n\n**What's there in the examples submission files ????**","metadata":{}},{"cell_type":"markdown","source":"#### 1. example_sample_submission.csv ","metadata":{}},{"cell_type":"code","source":"sdf=pd.read_csv(\"../input/tensorflow-great-barrier-reef/example_sample_submission.csv\")\nsdf","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:05.801257Z","iopub.execute_input":"2022-01-03T12:50:05.801697Z","iopub.status.idle":"2022-01-03T12:50:05.825392Z","shell.execute_reply.started":"2022-01-03T12:50:05.801662Z","shell.execute_reply":"2022-01-03T12:50:05.824135Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"so here for $0^{th}$ annotation :<br> \n`0.99` -- it's showing that there is bbox at given position and with given size confirming it's existence with CI of 99% or simply put it is probability of bbox being present there , <br>\n![image.png](attachment:b697bdcc-5315-46be-a44a-1a1e5b1512b3.png)<br>\n`123` -- is x co-ordnate it's  $b_x$,<br> \n`456` -- is y co-ordinate it's $b_y$,<br>\n`1` is weidth it's $b_w$'<br>\n`1` is height its $b_h$,\n\nthe image itself is solution but hard to implement 😂","metadata":{},"attachments":{"b697bdcc-5315-46be-a44a-1a1e5b1512b3.png":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAUcAAADxCAIAAADXz+LgAAAgAElEQVR4nO3de0BMaRsA8GemmaabRJRKSqWUsF2USi65lYpcInZZocvGrmWxLru5htYKS0QKuZZbLcqWWHbXUoSU1VJJRUWprWmmppnz/XFyvm5qmmbmzJze31/T2ztnnpkzz5zbe56XhmEYIAhCIXSyA0AQRMxQViMI1TDIDkA8ysvLT58+/fTp0z59+gwbNszZ2bl///5kB4Ug5KBR4Lj677//njlzZllZmbW1dUVFRW5ubnBw8ObNm8mOC0HIIffbajabPW/ePA6H8/jxY0tLSwAoLCzkcDhkx4UgpJH7rE5KSnr16tW+ffvwlAYAfX19ckNCEHLJ/dmy7OxsALCzsyM7EASRFXKf1Xp6egCQmZlJdiAIIivkPqtHjx4NAGfPniU7EASRFQqbNm0iO4Yu0dTULCwsvHz5cnZ2NovFKi4ujo2NLS4uJg6zEaS7kfuzZQAQGRmprq5+5syZCxcuAEDPnj137txJdlAIQhoqXK/GYRj2/PlzVVXVAQMGAACHwxk3btzWrVuHDh06derUhIQEHR0dsmNEEGmgTla39uDBAy8vLwMDg2+//dbb25vscBBESqic1QAwd+7cv//++8WLF0wmk+xYEERK5P4ceDtSUlJevnw5cuTI1atXkx0LgkiPlLbVNJq0dwpKS0tHjRqVlJSko6NjZ2e3ZcuWmTNnSuKFpP/WEKR9lM1qqaHwW0PklPSubNnb20vttRCkO5NeVqMNGoJIB9oD7yoKvzVETlH5HDiCdE9oO4MgVCP+4+qffvrp7NmzysrKYl+yXPjvv/9UVVUVFBTIDkTa6urq+Hy+iooK2YGQoLa2VkFBgcViSefl7t69234H8Wf1y5cvrayslixZIvYly4XZs2cHBwd3w1qIV65cefr06fr168kOhAShoaFmZmZeXl5kB9JI/FlNo9Hs7e0dHR3FvmS5oKSkZG1tbWJiQnYg0vbPP/+UlZV1z/Xer18/Y2Nj2Xnv6GwZglANymoEoRrR98BpNFqLFnQ6HUFkgejbagzDHj9+DACRkZEYhqGURhAZgfbAEYRqUFYjCNWgrEYQqkFZjSBUg7IaQagGZTWCUA3KagShGpTVCEI1KKsRhGpQViMI1aCsRhCqQVmNIFSDshpBqAZltezi8/mHDh1KS0vjcrmxsbH+/v6HDh0SCARkx4XIOpTVsovL5QYFBV28eNHDw8PHx+f48eNBQUH+/v5kx4XIOpTVsm7Xrl1v3rxJS0t79eqVqalpdHR0cXEx2UEhMg1ltaxTV1ePj48fMWKErq6uh4cHhmE5OTlkB4XINJTVsi4oKMjU1BR/jJcuVVRUJDUiRNahrJZ1TXM4JyeHwWBYW1uTGA8i+1BWy42SkpJjx47Z2tp2z/kxEOGhrJZ1T548SUlJiY6Otre3p9Fox44dIzsiRNZJb/5qRDTx8fHx8fGqqqouLi6rVq0aPHgw2REhsg5ltaxbt27d8uXLNTQ0mk7O9tVXXwUGBkZFRQ0ZMsTExOTly5cBAQEkBonIFJTVsk5RUVFbW7tFo7a2dnZ29u3bt3Nyctzc3MzMzEiJDZFN6LhaLllYWFy9etXBwaGgoCA9Pd3c3JzsiBAZgrbVsktZWTk9PV1XV7f1v8zNzefNm3fmzBkmk3ny5MlTp05JPzxEZrXMagzDzp8///fff9fX1w8dOtTW1tbW1paUyBA6nf6pD9/U1JRGo40bN65nz553795tPecZ0p01y+rq6movL6+bN28OHDhQS0vryJEjgwYNevbsGVnBIZ/CYrF4PB4ATJ48efLkyWSHg8iWZlm9du3amzdvhoWFrVixAgBqamqys7NJCkwMysvL2Ww2ACgqKvbr1w9vrK6u/vDhA/64b9++ysrK+OOioiL8JkdhOhcXF/P5/Bada2pqKioqeDzemzdv9PT0WndmMpk6OjpNO+OP+/TpQwwsabMzm80uLy9v3fnNmzcNDQ2d7cxgMIi9+qadNTU1VVVV2+lcW1v7/v371p3fvn3L4/HKy8tra2uJD7/9zgCgpqbWu3dvQCQB+6i+vl5RUdHJyQkTWos5MXH+/v4RERHCL0RyvL298fdobW1NNO7evZt47wkJCUS7uro63jh8+HCice/evUTnS5cuEe29evXCGy0tLYnG/fv3E53Pnz9PtPfp0wdvNDc3JxoPHjxIdD537hzRTpzuNjU1JRoPHz5MdD59+jTRTuSbsbEx0Xj06FGic0xMDNGur6+PNxoaGhKNTce0HDt2jGg3NDTEG/X19YnGmJgYovPRo0eJdmNjY7xRWVmZaDx9+jTR+fDhw0Q7Mabd19cXowrZ+c7j/n8O/MWLF/X19fb29m2kvpx78eIF2SEgjdC6kIL/Z7WOjg6NRnvy5AmJ0UhIdXU12SEgjTA0z7nk0Zp+ysOHD3/27Nnr16+Jg7T2PXny5LPPPouMjFyyZAnRGBAQYG1tLQtDnf7666/Xr19zOJzPPvuMuM+poaEBP64DABaLRac3/q5xuVz8o6DT6cQoLj6fX19f36nOQ4YM+fXXXy0sLIRfsqKiooKCgqQ719XV4ScOJNT5+PHj9+7dO3HiRPud09LSsrKylJWVjY2N7ezs2lhtckh2vvO4ZmfLtm7dOmvWrDFjxqxfv97IyCg3N/f58+ehoaFkBddFTk5OTk5OLRoZDAaD0cZVeiUlpdaNCgoKxEkvITvT6XQlJSUipcW45C52bjrgVBKdFRUVm36wn+psZ2dHmWSWWc2+31OnTj137lxwcLCvry8AMJlMdNUEQeROy63WjBkzZsyYUVpa+u7dOzMzMyaTSUpYCIKIrO1x4Nra2paWlnhK8/l8GxubgoICAMjPz580aZJUA+yCtWvX2tjY2NjYREVFkR0L0ujAgQP4StmyZQvZsVBWx3d3KCgozJ07d+fOnQBw+PDhGTNmSD4q8cjLy8vIyMjIyHj79i3ZsSCNioqK8JXy6tUrsmOhLKHu7ggMDDQ1NV29evWFCxfwkScIgsgsoe7EVFNTCwwMdHV1nTJlipqamqRjQhCkK4S9E/Prr7/esWNHUFCQRKMRr5MnT0ZHR0Nb12mSU3iTg3gSeVV+2iBXRaDVdtyTYgTegM2MHtTsjf8Vw3R0aHbCdcuWLT/88AMAoBOxkiNsVt++fdvd3V2+imaxWKw2r7sCAJ8PwJfMq9KUoJvOhMUAaPmp8lt9yIqKiqieuaR1nNV5eXk+Pj76+vrh4eFSCAhBkC7qOKuNjIzS0tKkEIrYffjwgcPhAIC6uno7pwMyLolz0zF12rQD+/cPGDBAjMuUC/Hx8U+eZG7cGAwA1jPqP9WturoaH5avoqKioaEhvfi6EypXOAoICDh//jwAbN26FT+Wa01HA6w+E+eHwGLmDLEQmJhQ+YNtU8bDqqLCYvzDZCnX13Ha7hYSEoKPQfb19cXPeiBih6oRyrR37wRsNrrJCekclNWyi83GtOy5P4XVkR0IImeovKNoaWn57t07ACAqeyCkMzY2Hjt2LACgaseSQ+WsDg4ODg4OJjsKpBk/Pz8/Pz+yo6A4Kmc1ZXC5WPwV3o1bfOvh9EA/Fh0dNiHtQlkt62o5mNsszu/PAOgQdZ3/4BEnOqKNagQIQqByVt+4cSM3NxcA7OzsrKysyA5HRD/HYkZ9IC1OUU+X7jiNeywF21os0NOT1+11Wlrao0ePAGDw4MFjxowhOxxqktcvhzCOHDkSGBgYGBh47do1smPpAgYknWKNsGXo6tKnu9AAg5x/JTTYVRouXbqErxSiwhkidlTOampYO5tuOqixlN8gIzoAoGHUSPtQVsu6pjn8/IUAaGBtReXjJqTrqPz9+PLLL0eOHAkArSuNyqOSEsH+K5itIaioyPFceR4eHlpaWgBgaWlJdiyUReWsdnd3d3d3JzuKrnqULUi5wXtdJFi7lw8AJw+2fW+pvBg1atSoUaPIjoLiqJzV1HAlDa6k8YABU4bD998wB5spkB0RIutQVssuVVUalqfC40FFhUBDg8Zi/X/HO+BrTpAfM/I4z9KcPsiY/iJXEOgn39twRIyonNWhoaF//PEHAMyfP3/OnDlkhyMiJhO0tVue1OynRct6Jki9jz3P5buPxwabys0G/MSJE/jtsRMnTly+fDnZ4VATlbP64cOH+JVq/JwZlViY0y9f5Ttb0VLuY/cfYdOnys21jH/++QdfKfg5M0QS5ObbgDRlMZge+wc2fqzClFG02LuYgYHcbKsRKaDytprC8HEpLmMYGuq0Ow95NDm+1IWIH5Wz+pdffgkJCQGAPn36kB2LmLFYNCxXBQAmT6JPniRPJXhXr169ePFiAOjZsyfZsVAWlbO6X79+/fr1IzsKpBlNTU1NTU2yo6A4dFyNIFRD8azGMAzDUDU/2YJWiqRROatnz55Np9PpdPq2bdvIjgVptHbtWnylLFq0iOxYKIvKWY0g3RPKagShGiqfA9fT0zMzMwMqXtmSX1paWvhK0dHRITsWyhI9q2kfhz4QtWBl7RTInj179uzZQ3YUSDMrV65cuXIl2VFQnOhZLWs5jCAIDh1XIwjVUPm4+uHDh8XFxQBgbm4+aNAgssNBAAD++eefFy9eAMCAAQM+++wzssOhJipvq0NDQ6dNmzZt2rTY2FiyY0EanThxAl8pv/zyC9mxUBaVsxpBuieJ7IFzudzTp09LYsmdUlBQgD948uRJi3ieZPYFGAUA5bXVp09fFuOLVldXJyQkdMO7Su7du5ebm4t/znU8NwAlAEhOTnld8KFpt+zsbPwB0ZkCXrx4wePx1NTUpPNyn3/+efsdJJLVdXV1iYmJklhypygoKAwePBgAKioqWsRTXGKBZ3U9jyPeUGtra2/fvt2jRw8xLlMu5Obmvnv3rvHDbHDBs/revXt5ublNu9XU1OArRSAQyMKXRCwKCwtramrq6qQ003iHWU0T+wWqgIAAa2vrgIAA8S5WvJKu86YE8QBARwPeZKiIcckmJibXr183MTER4zLlQlRU1N27d6OiogBAaUhtHQcA4E4M03mUPN3+LRpZ+85T+Rw4NdBaFTpBIwVkzeHDh8kOoRl0tkzW4fctXr58GQBSUlJQSiMdovK2+siRI+np6QDg5eVFgUk8qOHSpUtJSUkAMGrUqC+//JLscKiJyll948YNvPS0gYEBymoZkZaWdvToUQDg8/ntZDWXi1VVYZqadIa0vqECgeDevXsGBgZ6enrCP6u8vDwvL2/o0KFKSkqSi62z0B44IovOxNb3c+BKZ6buioqK7du3GxsbOzk5/fXXX0R7TU0Nl8tt/7mPHz+2s7NTV1cPCAh4+/athCMVFspqpFurq6tzc3P78ccfTU1N4+LivLy8iH/17t177dq17T99zJgxCQkJrq6u0dHR48ePr66ulnC8QqHyHvjmzZuXLVsGAIaGhmTHgjQKDAycMmUKAEhvoA6/6t+MR4W16oOGmmric5XRGMoqLPrHeNLS0mJjY2fPnt3ieQKBoMNlMxiMqVOnTp06NSUlxdXVdf78+ZcvX2592ULKqJzV5ubm5ubmZEchW16+fMlms4cPH96pZ/35559mZmZ9+/btegCGhoad+pEtKhLEXqj/Nxeb580YM7qzl74Fpfcj/BYGX3tRo6KE1dbWC/ALCCoeV0sT3NXokZGRx48f/+abb1qndGdNnDhx48aNGzduDA0NbbGFLysrq6ysNDU1bf2sjIyMK1euFBQUmJmZDR061MXFRSzH52gPvFvAMCw2NnbcuHGDBg1qWpuRw+EIs9M4bdo0LS0tJyenO3fuSDLMljKz+AaTuKsOCI5cw8b68pKu8zr1dP6H5JXz1mXrLr1T8KG6qvTOgXk6Kkbfnvjr8b1fnFXoABAWFtavX7+ff/656bO4XO6ECRMmTJjA5/MvXLgw4SNirOun/PDDD4MHD963bx+f3+x0wPHjx93c3Fr337Fjh52dXVhY2MuXL3fu3Onu7v7y5ctOvcFPQVndLYSEhPj4+BQXF4eGhh44cIBoX7RoET5+s323b99evnz58+fPJ0yYIM3EnreuYZ4TrfSuUuJBJmCweXcbWS0QCPhtEQgEpddjr7wxWxO+3klPGRQ0nAK2Lx3J/S25YNDQgep0yMnJef78+ciRI5nMZrsAdDrdzMzMzMyMRqNpamqafdThKGA6nT5q1KiSkpL79+93+NZu3bq1YcMGZ2fnsrKyO3fulJeXP3jwwMjIqFOfzycjEctSZJO/vz8+fcfevXvJjoVMV69e3bhxo7e397///rtmzRptbe3OLsHS0nLv3r25ubn6+vqzZs0qLCwUOZiQkBB8paxYsaLDzmMGw/EIZS0tupsrU7MH3C9qo8/KlSsZbVm82Pf1y0JeX0tLg49Jq6Br/Vn/0udZRTwMAHJzcwHA3t6+xQIVFRXDw8PDw8PpdPq4cePCPxowYECHAeNLy20+9L1NERERGIadOXOGxWIBAJ1Ot7GxUVERz+BlKh9XV1ZWlpaWAkBNTQ3ZsUgavzz30ZN8dh8TS+O++IEZjamkoqgABQUFn3/+uYmJCT5Cuys0NDQuXLjg6Og4ffr09PR00c4JVVdX4yulqqqqw857tysqfJzt00IPHr5uo4+fn9/48eNbt/fv36/u2rqWrRgm4DeeBMO/FeKtVImfesCXPHv27JKSEgAoLCx88+bN6NGj8T6hoaEODg7Z2dn6+voSKslI5azuJho+PNgVsCTkYpZARYlXW9vQeEZIcXFMydH5vaKiov77779Lly612IH84Ycf7t27l5WVVVlZOWHCBLxx1qxZgYGB7byWlZXVd999FxIScvv27bFjx0rqLX3EZDb+cGAYZBWBy5A2+rx//77NbaOysvIQEwPFd5lP83lOFiwAAH7xw8fFGgYDdRg0ANDX1weAhw8fLlmy5FMBdHZ87oMHD4glDx48GE/a+vr6Dx8+WFlZ4X3U1dUBQFdX9/nz5+/fv5dEAVwqZ7WKigr+CcrUuB9xqzq/ZvGWP3ptT85bOl6vMiN64fRV7ybsOfjNiH4DegBAfHw8g8FwdHRs8TRdXV0zM7OCgoLKykq8lC8It+EaPXp0SEhIQkKCaFmtpKSEr5RO7W2eOFX3oQbGOrRxwPjnn38eOXKkdfvMmTNdgudM1buwe+VP1sdWj9D8L3X/6gN/wZzTnj1oAADDhg1TVVVt5xi4f//++fn5wgcJAPfv32cwGHZ2dgCwZcsWvPGnn346fPjwvn37mvYcPXp0SkpKXFxcUFBQp15CKJi4+fv748cMsiwxqR4GsmEgW8eKLd4lGxsbv3jxQrzLxJrf3dEU/8PFKX1VPXZk8hsbGjL3TlXRnpNc1digqqpqZWX1qcX6+Pjo6up2KpIPHz7QaDRPT88W7UePHl20aBH+mGXBxj/eO3/Ud2rhhKjjXBjI/imMk5xSvz6YA0ZsZzc2j9fZxfBL/g73HKxJV2ApsxSY6oO+/Pn3jx8MhmHYggUL6HT6rVu32nzy1KlTVVVVY2NjHzx4sG3btvj4+PZfLCMjg8Viubu7t2gPDQ01MjJq0fj27VsTExNFRcWtW7feuHEjPj5+2bJlBQUFnX2HbaLytro74BblFbE13IYbftyKKRhYD9OuvvpPfv3E4Up8Pp/D4YjlOjOhR48eLBZL0qcq6HQAgDX7BQCCXmqwxJW2cwur82PC6dojg37NXpD/NLOwRsV4+FA9NYWm/961a9etW7d8fHweP37celTMvn37PDw85syZAwCqqqr48PVPqaysnDVrVo8ePZpeYvj4XtrYxejXr9/NmzcXLVq0bds2vNzCgAEDhDmDKAwqnwPvHlqfssIwaDwjpKCgoKOj8/Dhw08+mUYTZgRVU9nZ2VwuFz90lJyF81lYnkrNU+W3d5UqMlUiw5U1NRu/q/7LOJtCuAIBTJlVm5wixBVsutrA4Y6jnT5rkdIAoKWldfny5aqqKg8Pj2vXrrW4zmxoaJiVlfX69esHDx6UlZX5+PhwOJyRI0empKSUlJTY2dnhA78xDEtNTfX09Hz9+nVcXFzrMTarVq1q88hfX18/JSWlsrIyIyOjsrKyoKDAyMgoJibG3d2dz+dv27Zt06ZNQn9gzVB5W52bm1tRUQEA/fv3p+r8L0r6RgPUKrMe5wvchtEBAPivHmaWMAcY6jdeznF0dDx//nxubq6xsXHrp+vp6ZWVlXE4HGVlZSFfET8QbX2gLqSioiI8Gfr27dvhIDNVVZqqasufrZ+2KQ2bwEl7VGugR5s0sauFVmxsbGJiYvz9/T08PPr373/x4kX8qJigr69P/IQpKysfOHDAy8vLwMBg9erVOjo6z5498/T0zMvLU1NTi4iIGDduXGcDUFJSIk6kAcCCBQt+/fVXLy+vkpKSP//8U7Q3JWfbaj6ff+jQobS0NC6XGxsb6+/vf+jQoU9tbdatW2dnZ2dnZyfCRZ137wRsthzUJ6D3dPnCyyR1/3cH7xTxoL7wr/DvdqcO8vx8bO/G7dKCBQsA4FNX7IcPHy4QCNavX5+VlRUdHb1x48b2X66uri4iIkJJScnb21u0gA8cOICvFOJkUmdpaNDWBygkZcCKZYqiLaEFb2/vV69e7d+/f8KECR3WHrO1tXV2di4uLsbvA2loaHBycgoLC8vPz1+8eLFY4gkJCbl69aqfnx9+KVsUYjk6b0qiZ8vww7k1a9bglyjxUUGLFy9uszPxzdu6dWuLf7V/tqymRgAD2cFbOCJEKOWzZRiG8SrSQmYNV8XPCCn0sJqx82F5s9NKX3zxBQDExsa28Vweb9GiRfiBH41G++KLL9oPAy/NFR0d3fpfQp4t+/777/GV4uvr2/5rfcqbN3x9G/b6YM5nY9h1dQLRFiKy5ORkW1vbOXPmLF++XBLLb2hoGDt27MaNG3V1dfPy8kRbiJxtq3G7du168+ZNWlraq1evTE1No6Oj8Tk6uidGrxHrz2cU5T+6cT31cV7Rg4vfW/dudmB15MgRGxubJUuWHDhwoLKystlzGYyoqKjKysr09PTi4uKTJ09+6lVyc3NXrFhx+PDhr7/+2tfXV1JvpiMYBnMWc3/6nhGyWclyEG3l2g7ufxav0tLSoKCgs2fPRkVFpaSkXLx4UewvsWXLFnNz802bNm3bts3b25vH69zQd5xcHlerq6vHx8fjN8F4eHiEhYXl5OS0LmExevRoRUVFABg6dCgJUUoVXWPA0FGfGNGorKx8+fJlDw+Pr7/+evXq1Rs3bmxxU1GPHj1sbW3xx3w+387O7tKlSwYGBvn5+QEBAcnJyZ6enteuXWMwGAsWLAgLC+tKoFZWVnjhWwcHBxGeTqPBncTGC90nI4U9FyAu2tra+HRC0KSwuXht3rwZf+Dr6yvyr6dcZnVQUBBxXxteoxfP3haWLVuG318tMi4Xi7/Cu3GLbz2cHujHausKhXzQ19d//PhxQkLC7du32x/pqaCgMHfu3J07dx46dOjw4cMzZswAAHV19e3bt/v6+oowhryFOXPm4NeKEMmRy6xumsM5OTkMBsPa2lrsr1LLwdxmcX5/BkCHqOv8B4840RHS3jiIEY1G8/Lyalrr41MCAwNNTU1Xr1594cKFx48fA4D0J9lAdcu6Qm63PgAAUFJScuzYMVtbWxUVlYaGBtEOQj7l51js9TtIi1MsvqNkoAnHUrDi4s5d2pVTampqgYGBrq6uU6ZMkdosMy2gumVdIZdZ/eTJk5SUlOjoaHt7exqNduzYMQCIiYkhjklwcXFxwcHBwcHBIt4SzICkU6wRtgxdXfp0FxpgIJ0vmSz4+uuvCwsLJTFEOSUlBV8p8fHxYl+4CFDdMlkRHx8fHx+vqqrq4uKyatWqFvf9CwQC/FLNhQsX8MrBioqKxH1wwls7m246qPGq7yAjOgC/rYN3arp9+7a7u7swBRU6KzU1NTQ0FAB8fX2FORwQg+5Xt0wut9Xr1q0rKSkpLy//9ddfm6ZrcXGxo6PjgAEDzp492/VXaZrDz18IgAbWVnL5I9gpeXl5dnZ2J0+ebD2emRRFRYLde7kBX3Nu3xHh8EpQev/gVEtjcwdXL3eHgX16qOH6zkqqEQCA2OuWJSQk4L9ZTZWVlf37779tPisjI2Pz5s2LFi0KDQ1NTEzscIdfSHKZ1YqKitra2q1H3ly6dCkyMvLkyZNiyWpCSYlg/xXM1hBUVGgYBlyuHIw5E5mRkVFaWtrFixdlYbJeVLdMNJTa+Hz11VdDhgwRCAT4uL8VK1bMnDkTRL1e/ShbkHKD97pIsHYvHwBOHmQBQP4r/jz/unsp4pxGs31Nd+cmTpwIcj573rx58/Bhz8LU6Jq3ruELF9ruENbDDP6UIN7m3Tw315YDvwUCQZsfCI1Gw+uW7bq23kmPBaDsFLB96UXH08kFIQscVaCxbpmXl1ebdcsA4ObNm3jdMrxdyLplR48evX//fofj5PG6ZWPGjLl+/TqLxRIIBI8ePRJX3TI5y2plZeX09HRdXd02/8tofhnEwcFBtKEOuCtpcCWNBwyYMhy+/4Y52KzZHT8CAUjn8rVc53Brw4YNGzZsmJCd8bplCgrg5krX7MH7VN2yFgUJcAsXLggwKm5dtyzsdlYRDzNl0tqvWwYAhw8fHjduXKeK3tnb2x89ejQ3N7fDrG6zbpnwL9Q+OctqOp1OjIKSHFVVGpanwuNBRYVAQ4PGYjU7+VHHg3mLOIkPsC8m0A6EyfEVbNmH6paJRs6yuh2LFi3CHwwdOvS3337r+gKZTNDWbmNz/Pg1fBdAP3ZIUcWKc6Dd0ZM0Gk3kLW2L55aVCfLyBSbG9D595PJUiGhQ3TLRUCerW/v++++Tk5MBYOnSpe2suc6yMYQv5rIAQEla+ZX+oMHu83rgAbAg87ziUEs5Xmv79++Pjo4GAC8vrw5v/CSgumWdIsffjw7l5+fjAx7xHSFxYbSsqNGeWbNmdf0VF35bDw0AACQuYZ4AABOzSURBVFAPAd/V3/1NjtdacXExvlKalgr4lMTfeG/eCH7/g7/9lMB5MCxf1saozA0bNmzYsOETCxD8fGqHn2+wg34Ii9HQwDKat/3M1hk6+G+DqqrqzJkzT5069fvvv7dZWXH48OGpqalxcXHGxsbXr1+3tLScNm1aO9E+evTozp07kydP1tLS6vCtLVmy5MSJEytWrKioqHBwcKipqblx48bq1auFqTreITn+fpCOLsRYg6YjEEXGrgPA9wQxqJXqrYekQXXLukRMN3v/n+zUGBW5akJXNK2a0JWPt+lzj8VwwYgNA9lgxI67UNfVECVDElUTamoEb9/ym7Y0NGCWzuxXrxowDMvLb3Dx7Orqe/DggZKSko2NzdWrVxsaGlp3wOuWsdlsDMP+/fdfe3t7Pp+PYdipU6c2btyIYZhAILhx48aoUaMYDMbNmzc7GwCHw8HrluF/+vv74/UtBAKBvb19RUWFCG+KylldV1fHZrPZbDavVclZ+cpqDMNSb9b/vIcjciFeKRAyq+vr6/GVUlcn4s/TT2Ec/2W1GIatXl978DC3KzHj4uLiNDQ0AKB///73799vv7O7u/u5c+cwDBs5cuSzZ8+ys7Pxi8xqampHjx7tejBpaWn4mIukpKQ5c+aIthAq74ErKiq2ed+11GBduNTc4rku45gu47paeU8WMJnMFqM+OusrP5aOA+f7fP7p61jObTGsX29v70mTJp08efLhw4cd1i3bsGGDv7+/iYmJioqKubl5Zmamk5PTsmXL5s+fL5az2SNGjNDT07t48eLJkydXrVol2kKonNUIJamp0b7zoY/zrps+hqamJp77KHr27ClkgQ0HB4e+ffvihSUAYNiwYTExMWKJgbBhwwZfX191dXVnZ2fRlkDli58VFRVFRUVFRUUycn8cAgD//fcfvlI+fPgg8kKWB7FeV8GyAHJ2xFatWsXlcqdOnSqh5Y8ePVpTU7P9Cc/aR+WsDgwMxIs5tzmiECHF9u3b8ZXy3XffibyQ23d4XnbQYgyv1CQnJ69evbrp8OSAgIDDhw+La/kFBQVsNnvhwoUiL4HKWS2/BALB3bt3O1s4tby8PD09XVx388mmvHy+1dja47ENh3aTUFEoLi7OycmJz+dLZMo7AACYNGmSn59fRESE6MXA0XG1rKmoqIiIiIiMjHz16lXTW/lramoYDEb7xbEeP348YcIEJpPp6+u7adMmuZ6u5FN1y4wGKjz6XXo3zLUwe/bsrt+J3T58NGQXUXlbPXToUBcXFxcXl4EDB5Idi1AoWW2nBRMTE3ylWFhYtNNNmnXLhCHGHWwpoPK2+scff/zxxx/JjqK57ldtp4UlS5aIcUy+1Jw4cQKft0QuUHlbLWNkotoOIoKlS5fm5OT4+PiIPDeYlKGslhIZqbYjd7pSt6xTcy3iMjMzd+7cGRAQcOLECWIgUHh4uJmZ2blz54KDg0V8G9JF5T3wlJQUvBCUvb29JKYB6BRZrrYjTffv38/IyAAAc3PzNu+Uaiozi//FpgZBPYAAjiTyEsOhdYWjdnC53KCgoDVr1qxfvz41NZXJZEZGRj58+PBT92lER0cvXrxYQ0OjV69eR44cuXHjBjHxWFeGCUoflbfVkZGRQUFBQUFBiYmJZMfS8PplYetqO6XPs4p4GAC0X20nPDycTqePGzcu/CNh7tfDl9ZmRQESXb58GV8pwgzJmreuYZ4TrfSuUuJBJmCweXcbm2uBQMBvC7FNFnKuxQcPHgQEBLi6uhYXF798+dLPz+/UqVOPHj3C/zt27Fhvb295OSNA5ayWLa1PWkmx2o6cwuuWaWnR3VyZmj3gU3XLGG0hZpPG51ocMWKErq6uh4cHhmE5OTmtl5OQkNDQ0BAYGFhVVVVaWjpp0iQAIGoq7Nix48iRI4cOHYqIiPjyyy/FW8RW7Ki8By5LGAakVtuRU12rW9YffyDkXIv37t0DgBYTDxQV/f+HpFevXgDw4cMHVVXVT9X3lhFUzuqFCxfih5SycGCp7UpmtR3Z4enpidcnGDKkrSpkzXWxbhn+QMi5Fvv27ctkMl+8eNH01EbracZcXFwMDQ2J+W5lE5WzesqUKVOmTCE7ikZ0jQkyW21HmpycnJycnDr7LNHqlrWoG910rsXW/Z2dnc+ePZuZmenp6dlOJPjZiq7P+CtZXb/PuwXZqZrQDulUTWgDvzrv8V+3/3xUVN2y7EZpaam+vr62tvbbt29bPy8/P5/YuKmqqp49e7adGD58+GBkZNSnT5/8/HyR3oQohKyaIKSo41wYyP4pjJOcUr8+mANGbGc3dqviFx3Azyl4eXklJydHRUUNGDCgZ8+e//zzT5udKyoqzM3NNTQ0YmJinj17du3aNR8fn8jISGFeSNa+8+hsmXTR1QYOdxzt9FmLAloAoKWldfny5aqqKg8Pj2vXrrW4zmxoaJiVlYVX2ykrK/Px8eFwOCNHjkxJSSkpKbGzs8OnWcUwLDU11dPT8/Xr13FxcYaGhtJ7a2JF1C2b5M87dFmwxJV2+aSSaHNZx8fHT5o06Ztvvhk+fPivv/76qSkBe/XqlZSUZG5uvnDhQgsLi/nz53O5XNFmfSGf2H8nZOd3a8eOHW5ubm5ubq23bKRtqzvSqWo76enpenp6jo6OcXFxGIaJvdpOpwi5rT527Bi+Uvbs2dPhMlvXLXuZ2zBvUW1BQYOBLZvPx2Z8Xtvu02vg41yLXG6zWkgODg4YhllbW2dmZm7ZsuXp06fEv8rLy58+fYpP9CMk2fnO46h8XJ2RkZGUlASycbZMSJ2qtmNra+vs7Pz333/jZ24bGhrEW21HEp4/f46vFGFm51NVpamqNrskOEBf4X42duNWQ8F/cO8+r6q64+sC+FyLrRv/+uuv3Nzcq1evZmRkrFy5kvhX7969e/fuLdSbkVVUzmo5JXy1HXzw3MiRI1evXr13715JVNuRNUwmMOiQeEOwfQn9yPEGHS0R712xsLAIDw9fu3btlStX1NTUVFVVxRsnudBxtbwqLS0NCgo6e/ZsVFRUSkrKxYsXyY5ISgYbwtOX2JKFiidSMcvB7WU1Pteiv79/63+Zm5ufP3/e19eXzWYzRDtel2FUez9N7d+/f8eOHQCgqalJdizip62tTVw17fBmD9mxZs0aPz8/+DjdlAgsTem6WljfvnRbA7Awa2+z1M5cixYWFmZmZtra2m5ubvX19aJFIrOonNXa2tqyfl2x++n6Ueu2TY0FYdJTRS+KMn78+KysLADAf/cpBu2BIwjVUDmrMQzDb9/B5Oo2OmoTcqVwuVhpqaChQWpxUQqVs3rOnDn47TshISFkx4I0WrduXYt7qtqE6pZ1BZWzGkHE5cSJE2SH0AkoqxGkA3JXt4zK58D79+9vbm4OH+sHILJAW1sbXym6urrC9C8qEsReqP83F5vnzRgzunPT7vH5/CNHjtjY2AwbNiwhISE1NdXKyiogIKDNCaVxmZmZiYmJ+fn5jo6OCxYswCu0hoeHP3r06Ny5c516dRJROavDwsLCwsLIjgJpZsWKFcLPvS6DdcsEAkE7PwoyQtbjQ7ozGaxbtmfPHsm9X3FBWY3ILhmsW1ZfX799+/aDBw9K6j2LA5X3wNPT0/FfZQsLC6JyFUKuZ8+e4UW/DAwMrKys2u8sg3XLmEzmqlWrdu/e3X7k5KJyVu/atev8+fMAsHXr1h9++IHscBAAgJiYGHxGEV9f3+jo6PY7y2DdMrw8k4zf20vlrEYoo2ndMi4XYzJpxDZctLplDQ0NGIY1TWAh65bhy2yxZFmDshqRXYm/8d68Efz+B3/7KYHzYFi+TAkAlizjLpzLmDC+MSE3bNiwYcOGNp/OZrMB4MmTJykpKYWFhZs3b6bRaMeOHQOAmJiYvLy8bdu2EZ19fHz279+/YMGCX375xdbWNj8//+TJk+PHj5eXyv5NUTmrvby88DluZXx/qVsZO3YsfpVoxIgR7XQj6pYBCHqpwRJX2s4trKb3QeNXmoSZ8DM+Pj4+Pl5VVdXFxWXVqlUt6pYRV6rwumVz585duHChQCDo3bv36NGj5bRuGZWzet68efPmzSM7CqQZV1dXV1fXDrstnM9aOB/YbKy6GuvXr+WVmtTf+QHreex6+P0ca7BZy7qOLaxbt2758uUaGhosFqtpe3FxsaOj4+vXr3ft2jV37lwAMDAwuHv3bkVFxZs3b4YMGSJrkwQLj8pZjci71nXLcIl3BBnJyrv31V1L4nWY1W3WLQOAS5cu3b17t6ysbM+ePXhW41DdMgQhQeh6Zs+etD6atBq26PfYfvXVV0OGDBEIBB1WfZQ7VM7qiIiI9PR0AJg+fbqHhwfZ4SAAABcvXsSnKHV2dl64cKFoCxGy0Bhet+xTA86pV66MQNk3BgA3b97Er1cPHDgQZbWMSE9Pxy9TYxgmWlYTR7sdHva2U7eM2mhirxMSEBAwaNCgpoNyyJKYmIjX63NwcGgxidyrApOEJ4sAAOjvl3uI8w6Q48ePe3l54ZX6u5WsrKy3b99OnDgRAPZdWQdYDwCYZRuhp9tsRNiff/758OFDALCwsMA7U0BqaqqWlpbUTpjv3bu3/Q4S2VbT6XRZmAuGmCRNQ0OjRTy13MbTJwwFpnhDZTAYenp63fDez5KSkurq6sYPk64AfACAfv36GRo2O4P99OlT/IGamposfEnEokePHpqamrLzdiSyrba2tg4ICBDvYkWQk5Pz7t07ADAwMGgxjXPSdd6UIB4A6GjAmwzRS1W2ZmJicv36dXy8cbcSFRV19+7dqKgoAFAaUlvHAQC4E8N0HtXs3smCgoLCwkIA0NbWHjRoEBmRip/sfOdxVD6uNjMzMzMzIzsKpBkDAwMDAwOyo6A4dCcmglANymoEoRoqZ7Wfn5+WlpaWlpZc1K/oJrZt24avlG+//ZbsWCiLysfVVVVV+Nky/N4dRBbU1NTgK+W///4jOxbKovK2GkG6Jypvq9XU1PDRIERZDIR0Kioq+Eqh2JTRMoXKWR0dHd1hDR1EyoKDg4ODg8mOguLQHjiCUA3KagShGirvgb98+bK8vBwA9PX1hZz/BZG0wsLCN2/eAICWlhZefwoROypvq9evXz9y5MiRI0eio2vZER4ejq+UrVu3kh0LZVE5qxGke0JZjSBUQ+Xj6jFjxigpKQHAsGHDyI4FaWRtbT1//nz4OAkGIglUzuqlS5cuXbqU7CiQZmbPnj179myyo6A4tAeOIFSDshpBqIbKe+CxsbF4layJEyeOGTOG7HAQAIDk5OQ7d+4AgI2NzfTp08kOh5qonNUXL17EKwcrKSmhrJYRN2/eJGa6RVktIWgPHEGoBmU1glANlffAV65c6e3tDQCWlpZkx4I0+vzzz21sbAAADQKXHCpnNT7euP0+byuBNqhWnK/KTxvkqgg0sS5TLgi8AZsZjX+Y/E/2Gjp0qJxOCi1HqJzVwvr0V1AUNCUQiHWBcoMBIO4PExEJOq5GEKqh8rZ6zZo1v/32GwAsW7bMz8+v6b9GOTEyLnU0paJIpk6bdmD//gEDBkhi4bIsPj7+yZPMjRubVS8yHdRyyvh9+/bhN8ZOnz5906ZN0ouvO6FyVr969SozMxMASktLW/yrRw+a1WcSee8sZs4QC4GJCZU/2DZlPKwqKizu8FN9+/YtvlLwc2aIJKA9cAShGpTVCEI1VN5RPH36dExMDAAwmcwOOyPSsW3bNvxwmsGg8nePXFT+ZJlMJspnWcNgMFA+SxraA0cQqqHyr2Z5eTk+b56Ghoa6ujrZ4SAAAFVVVVVVVQCgpqbWu3dvssOhJipvq7/66isDAwMDA4OePXsSjadPn7b+CL/RFzdmzBi8ce7cuUTj2bNnic63bt0i2seNG4c3zpkzh2iMi4uztrYuLCycNm1aamoq0T5+/Hi886xZs4jGCxcuEEtOTk4m2idNmoQ3zpgxg2i8dOkS0fn69etEu6urK944bdo0ojEhIYHonJiYSLS7u7vjjZ6enkTjlStXiM5Xrlwh2j09PfFGd3d3ojExMZHonJCQQLRPmzbN2tp68+bN+OgAXFJSEtH50qVLRLuGhga+UlatWgWIZFB5W92m0tLSR48e4Y8rKyuJ9sePH+NzrwoE/x/wWVZW1mbnJ0+efPjwAQB4PF7rzs+ePcP/i8vMzHz//j0AcLlcovHdu3fEklt0xq+uN52d9/3790TniooKov3p06d4xfyms8Y27YxPckB0LiwsbPFy5eXlbXbOysp69eoVvrQ2Ozdtz87Ozs3NheazFFZUVLTZGZECKm+rCdbW1mSHgDQyNTUlOwTqo/K2evny5Xi1jabHb+7u7jo6OvjjpsOboqOj6+vrAaBXr15Eo5ubm5aWFv54xIgRRPvRo0fr6uoAAJ+0FTd58uQzZ858++23a9eutbe3J9qPHDmCb6WbHghMnDjxzJkz+GMHBweiPSIigsPhAEDTEwHjx48nOjs6OhLtBw8erK2tBYAePXoQjePGjWtzyQcOHMC3/2pqakTjmDFjiM5N72/75ZdfampqoPl8tM7OzkTnpm9w79691dXVt27dysvLIxpHjRpFdLazsyPaw8LC8D0LY2NjQCSDhmGYeJcYEBBgbW0dEBAg3sXKCxMTk+vXr5uYmJAdiLRFRUXdvXs3KiqK7EBIIGvf+W6xB44g3QrKagShGpTVCEI14j+uRhCEXGhbjSBUg7IaQagGZTWCUA3KagShmv8BoUiGh6178q4AAAAASUVORK5CYII="}}},{"cell_type":"markdown","source":"#### 2. example_test.npy","metadata":{}},{"cell_type":"code","source":"sample_test = np.load('../input/tensorflow-great-barrier-reef/example_test.npy') \n#sample_test","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:05.832133Z","iopub.execute_input":"2022-01-03T12:50:05.832711Z","iopub.status.idle":"2022-01-03T12:50:06.109567Z","shell.execute_reply.started":"2022-01-03T12:50:05.832672Z","shell.execute_reply":"2022-01-03T12:50:06.108651Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 💥Preprocessing and feature engineering","metadata":{}},{"cell_type":"code","source":"# Change the type of 'annotations' from str to list\ntdf[\"annotations\"] = tdf[\"annotations\"].apply(ast.literal_eval) # str -> list","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:06.111516Z","iopub.execute_input":"2022-01-03T12:50:06.112274Z","iopub.status.idle":"2022-01-03T12:50:06.48254Z","shell.execute_reply.started":"2022-01-03T12:50:06.112184Z","shell.execute_reply":"2022-01-03T12:50:06.481797Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"`Note:` our train.csv only possess video_id , sequence, video frame , image sequence, image id but actual images are in video_0,1,2 folders we need to combine them , one way is copying path of images as per video_id ... thanks to Yoshi_k he has done it for us by writting below code    ","metadata":{}},{"cell_type":"markdown","source":"#### 🔅Adding image path to dataset","metadata":{}},{"cell_type":"code","source":"\n# Add columns of image path and number of bboxes, and the difference.\ntdf['image_path'] = main_path + '/train_images/video_'+ tdf['video_id'].astype(str) + '/' + tdf['video_frame'].astype(str) + \".jpg\"\ntdf['num_bboxes'] = tdf['annotations'].apply(lambda x: len(x))\ntdf.head()","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:06.484218Z","iopub.execute_input":"2022-01-03T12:50:06.484757Z","iopub.status.idle":"2022-01-03T12:50:06.597272Z","shell.execute_reply.started":"2022-01-03T12:50:06.484708Z","shell.execute_reply":"2022-01-03T12:50:06.596306Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### 🔅 new features and their insights ","metadata":{}},{"cell_type":"code","source":"tdf.sequence_frame.unique()","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:06.599844Z","iopub.execute_input":"2022-01-03T12:50:06.600156Z","iopub.status.idle":"2022-01-03T12:50:06.606909Z","shell.execute_reply.started":"2022-01-03T12:50:06.600114Z","shell.execute_reply":"2022-01-03T12:50:06.60613Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"a=tdf.num_bboxes.unique()\n#display(a)\ntdf['num_bboxes'][tdf['num_bboxes']==0].count()","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:06.608018Z","iopub.execute_input":"2022-01-03T12:50:06.60867Z","iopub.status.idle":"2022-01-03T12:50:06.620858Z","shell.execute_reply.started":"2022-01-03T12:50:06.608636Z","shell.execute_reply":"2022-01-03T12:50:06.619949Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"fig ,ax= plt.subplots(3,1,figsize=(20,30))\nsns.set_style('darkgrid')\nsns.scatterplot(data=tdf , y='num_bboxes',x='video_frame' , hue ='video_id',ax=ax[0] ,size='num_bboxes', palette=['r','g','b'])\nsns.scatterplot(data=tdf , y='num_bboxes',x='sequence_frame' , hue ='video_id',ax=ax[1] ,size='num_bboxes', palette=['r','g','b'])\nsns.scatterplot(data=tdf , y='num_bboxes',x='sequence' , hue ='video_id',ax=ax[2] ,size='num_bboxes', palette=['r','g','b'])","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:06.621981Z","iopub.execute_input":"2022-01-03T12:50:06.622397Z","iopub.status.idle":"2022-01-03T12:50:11.544295Z","shell.execute_reply.started":"2022-01-03T12:50:06.622363Z","shell.execute_reply":"2022-01-03T12:50:11.543653Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"so I tried to get more insight about how the annotations are scattered among the data as per sequence and as per video frame :\nconclusion is  Video_id 1,2 has more annotations per frame than video_id 0 , there are upto 18 annotations in 1 and 2 while 0 has only upto 5 <br>\nwhat could be the use of this annotation scattering ? are they random ? --- Umm video_id 1 and 2 are normely distributed annotations while video 0 has damped annotations ... what is the meaning of this then ...  it is possible that video_id 0 has some annotations which are wrongly placed or something like outside the grid annnotations or someone disturbed annotations of  video 0 or someone trimmed the data of video 0. but simply putting we don't know yet how this happened , it could be just wrong interpretation as well.","metadata":{}},{"cell_type":"markdown","source":"# 💥 Visualizing Image Data ","metadata":{}},{"cell_type":"code","source":"tdf.head(2)","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:11.545515Z","iopub.execute_input":"2022-01-03T12:50:11.545899Z","iopub.status.idle":"2022-01-03T12:50:11.557479Z","shell.execute_reply.started":"2022-01-03T12:50:11.545869Z","shell.execute_reply":"2022-01-03T12:50:11.556725Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### 💢converting annotations from [{'x': 666, 'y': 45, 'width': 32, 'height': 33}] to --> (666, 45, 32, 33)","metadata":{}},{"cell_type":"markdown","source":" ok .. now we want to get values of x,y,width and heigth of each rectangle annotation in each image  <br>\n `note:` there are upto 18 rectangle (num_bboxes) annotations in each image , so we need upto  18 x,  y,  width and  height values each","metadata":{}},{"cell_type":"code","source":"# splitting the annotation list into suitable parts \na= tdf.annotations.loc[134]\nprint(a ,type(a))\nprint(a[0] , type(a[0]))\nc= tdf.annotations[tdf.num_bboxes==18]\nprint(\"c :\\n\" ,c)","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:11.558941Z","iopub.execute_input":"2022-01-03T12:50:11.559262Z","iopub.status.idle":"2022-01-03T12:50:11.583201Z","shell.execute_reply.started":"2022-01-03T12:50:11.559221Z","shell.execute_reply":"2022-01-03T12:50:11.58265Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"tdf.annotations.loc[12679]","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:11.584507Z","iopub.execute_input":"2022-01-03T12:50:11.584915Z","iopub.status.idle":"2022-01-03T12:50:11.59287Z","shell.execute_reply.started":"2022-01-03T12:50:11.584884Z","shell.execute_reply":"2022-01-03T12:50:11.592176Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"we want to convert these values into one list ","metadata":{}},{"cell_type":"code","source":"\ndef access_annotation(i ,df):  # i = index number/ row number / loc value ,df = data\n    a= df.annotations.loc[i] #annotation number i \n    # accessing dict values\n    x=[];  y=[]; width =[] ;height =[]\n    for j in range(0,df.num_bboxes.loc[i]):\n        \n        x.append(a[j].get('x'))\n        y.append(a[j].get('y'))\n        width.append(a[j].get('width'))\n        height.append(a[j].get('height'))\n    \n    return x,y,width,height\nx,y,width,height=access_annotation(12679,tdf)\nprint(\"values of annotations :\\nx :\",x,\"\\ny : \",y,\"\\nwidth : \",width,\"\\nheight: \" ,height)\n","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:11.594981Z","iopub.execute_input":"2022-01-03T12:50:11.59542Z","iopub.status.idle":"2022-01-03T12:50:11.606096Z","shell.execute_reply.started":"2022-01-03T12:50:11.595385Z","shell.execute_reply":"2022-01-03T12:50:11.605476Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"ok now we got dimensions of each annotation , now let's plot it","metadata":{}},{"cell_type":"markdown","source":"#### 💢 plotting the images with annotations","metadata":{}},{"cell_type":"code","source":"# plot function \ndef plot_fig_with_annotation(i,df,figsize):\n    \n    fig ,ax = plt.subplots(1,1,figsize =figsize) \n    img = cv2.imread(df.image_path.loc[i])  # reading image\n    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) # adding color\n    \n    if (df.annotations.loc[i]!=[]):\n        x,y,width,height = access_annotation(i,df)\n        for j in range(0,df.num_bboxes.loc[i]):\n           # plotting rectangles \n            cv2.rectangle(img,\n                  pt1=(x[j], y[j]),\n                  pt2=(x[j]+height[j], y[j]+width[j]),\n                  color=(0,299,0),\n                  thickness=3)\n            # adding text to each annatotation \n            ax.text(x[j],y[j]-5 ,\"label \" + str(j+1) ,color = 'black')  # adds text label at top of image\n        ax.set_axis_off()\n        ax.imshow(img)\n\n            \n        print(\"THERE EXISTS ANNATATIONS FOR BOTS IN THIS IMAGE AS SHOWN IN FIG\")\n    else :\n        plt.imshow(img)\n        print(\"THERE ARE NO ANNOTATIONS IN THIS IMAGE\")\n        \nplot_fig_with_annotation(12679,tdf,(20,20)) #try value of i = 10 , 100, 5474 , 9291 , 12679","metadata":{"execution":{"iopub.status.busy":"2022-01-03T12:50:11.607366Z","iopub.execute_input":"2022-01-03T12:50:11.60792Z","iopub.status.idle":"2022-01-03T12:50:12.642683Z","shell.execute_reply.started":"2022-01-03T12:50:11.60788Z","shell.execute_reply":"2022-01-03T12:50:12.641994Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"interesting, isn't it ?😯 upvote if you like the simplicity in code 😉","metadata":{}},{"cell_type":"markdown","source":"## 🍀 Conclusion of DATA analysis :\n1. we need to implement Object Detection algorithm \n2. model should be less heavy to train for creating real time application\n3. Data augmentation is sufficient but not necessary\n4. Transfer learning is to be used  training entire data is not feasible \n5. we should be using YOLO , RCNN ,or something like that...\n\n","metadata":{}}]}