{"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":"![icevision-logo-slogan.png](attachment:a0b2546f-e0e5-4c7b-830e-cf28d8763345.png)","metadata":{},"attachments":{"a0b2546f-e0e5-4c7b-830e-cf28d8763345.png":{"image/png":"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"}}},{"cell_type":"markdown","source":"# 🧊 Loading icevision","metadata":{}},{"cell_type":"code","source":"!pip install icevision[all] > /dev/null","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:37:53.763254Z","iopub.execute_input":"2022-01-10T17:37:53.763981Z","iopub.status.idle":"2022-01-10T17:38:47.498166Z","shell.execute_reply.started":"2022-01-10T17:37:53.763943Z","shell.execute_reply":"2022-01-10T17:38:47.497295Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 📚 Importing Libraries","metadata":{}},{"cell_type":"code","source":"import numpy as np \nimport pandas as pd \nimport matplotlib.pyplot as plt\nimport matplotlib.patches as patches\n%matplotlib inline\nfrom PIL import Image\nimport glob\nfrom tqdm.notebook import tqdm\nfrom icevision.all import *\nimport ast","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:06.566713Z","iopub.execute_input":"2022-01-10T17:39:06.567329Z","iopub.status.idle":"2022-01-10T17:39:11.942280Z","shell.execute_reply.started":"2022-01-10T17:39:06.567286Z","shell.execute_reply":"2022-01-10T17:39:11.941517Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# ⚙️ Parse the dataset","metadata":{}},{"cell_type":"code","source":"WIDTH = 1280\nHEIGHT= 720\n\ndef get_bbox(annots):\n    bboxes = [list(annot.values()) for annot in annots]\n    return bboxes","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:11.944017Z","iopub.execute_input":"2022-01-10T17:39:11.944252Z","iopub.status.idle":"2022-01-10T17:39:11.951473Z","shell.execute_reply.started":"2022-01-10T17:39:11.944219Z","shell.execute_reply":"2022-01-10T17:39:11.949759Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"PATH = \"../input/tensorflow-great-barrier-reef/train_images/**\"\nimages = glob.glob(PATH + '/*.jpg')\ndataset = pd.read_csv('../input/tensorflow-great-barrier-reef/train.csv')\ndataset[\"num_bbox\"] = dataset['annotations'].apply(lambda x: str.count(x, 'x'))\ndataset = dataset[dataset[\"num_bbox\"]>0]\ndataset['annotations'] = dataset['annotations'].apply(lambda x: ast.literal_eval(x))\ndataset['bboxes'] = dataset.annotations.apply(get_bbox)\ndataset['video_frame'] = dataset['video_frame'].apply(lambda x: str(x)+ '.jpg')\ndataset[\"video_id\"] = dataset['video_id'].apply(lambda x: 'video_' + str(x))\ndataset['paths'] = '../input/tensorflow-great-barrier-reef/train_images/'+ dataset['video_id'] + '/' + dataset['video_frame'] \ndataset['label'] = 'Fish'","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:11.954286Z","iopub.execute_input":"2022-01-10T17:39:11.956866Z","iopub.status.idle":"2022-01-10T17:39:12.982207Z","shell.execute_reply.started":"2022-01-10T17:39:11.956816Z","shell.execute_reply":"2022-01-10T17:39:12.981469Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def show_sample_image(n=2):\n    img_index = list(np.round((np.random.random(n))*len(dataset)))\n    fig,ax = plt.subplots(1,n,figsize=(10,10))\n    i = 0\n    for idx in img_index:\n        img = dataset.paths.values[int(idx)]\n        img = Image.open(img)\n        if n == 1:ax.imshow(img)\n        else:ax[i].imshow(img)\n            \n        annot = dataset.bboxes.values[int(idx)]\n        for a in range(len(annot)):\n            xmin = annot[a][0]\n            ymin = annot[a][1]\n            width = annot[a][2]\n            height = annot[a][3]\n            rect = patches.Rectangle((xmin, ymin), width, height, linewidth=3, edgecolor='r', facecolor='none')\n            if n==1:ax.add_patch(rect)\n            else:ax[i].add_patch(rect)\n        i = i+1\n    plt.show()\n    return None","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:12.984325Z","iopub.execute_input":"2022-01-10T17:39:12.984728Z","iopub.status.idle":"2022-01-10T17:39:12.993486Z","shell.execute_reply.started":"2022-01-10T17:39:12.984692Z","shell.execute_reply":"2022-01-10T17:39:12.992752Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"show_sample_image(n=1)","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:12.996009Z","iopub.execute_input":"2022-01-10T17:39:12.996779Z","iopub.status.idle":"2022-01-10T17:39:13.489211Z","shell.execute_reply.started":"2022-01-10T17:39:12.996734Z","shell.execute_reply":"2022-01-10T17:39:13.487700Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"training_df = pd.DataFrame(columns = ['img','xmin','ymin','label','width','height'])\n\nfor i in tqdm(range(len(dataset))):\n    annot = dataset.iloc[i].bboxes\n    annot_array = np.array(annot)\n    label = np.array([dataset.iloc[i].label] * annot_array.shape[0])\n    xmin = annot_array[:,0]\n    ymin = annot_array[:,1]\n    width = annot_array[:,2]\n    height = annot_array[:,3]\n    img = str(dataset.iloc[i].paths)\n    inter_df = pd.DataFrame({'img':img,'label':label,'xmin':xmin,'ymin':ymin,'width':width,'height':height})\n    training_df = training_df.append(inter_df)","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:13.490177Z","iopub.execute_input":"2022-01-10T17:39:13.490448Z","iopub.status.idle":"2022-01-10T17:39:28.675012Z","shell.execute_reply.started":"2022-01-10T17:39:13.490406Z","shell.execute_reply":"2022-01-10T17:39:28.674187Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"template_record = ObjectDetectionRecord()","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:28.677318Z","iopub.execute_input":"2022-01-10T17:39:28.677707Z","iopub.status.idle":"2022-01-10T17:39:28.684291Z","shell.execute_reply.started":"2022-01-10T17:39:28.677663Z","shell.execute_reply":"2022-01-10T17:39:28.681659Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class MyParser(Parser):\n    def __init__(self, template_record):\n        super().__init__(template_record=template_record)\n\n        self.df = training_df\n        self.class_map = ClassMap(list(self.df['label'].unique()))\n\n    def __iter__(self) -> Any:\n        for o in self.df.itertuples():\n            yield o\n\n    def __len__(self) -> int:\n        return len(self.df)\n\n    def record_id(self, o) -> Hashable:\n        return o.img\n\n    def parse_fields(self, o, record, is_new):\n        if is_new:\n            record.set_filepath(o.img)\n            record.set_img_size(ImgSize(width=WIDTH, height=HEIGHT))\n            record.detection.set_class_map(self.class_map)\n            \n        record.detection.add_bboxes([BBox.from_xyxy(o.xmin, o.ymin, o.xmin + o.width, o.ymin + o.height)])\n        record.detection.add_labels([o.label])","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:28.685542Z","iopub.execute_input":"2022-01-10T17:39:28.685750Z","iopub.status.idle":"2022-01-10T17:39:28.699111Z","shell.execute_reply.started":"2022-01-10T17:39:28.685725Z","shell.execute_reply":"2022-01-10T17:39:28.698235Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"parser = MyParser(template_record)","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:28.702189Z","iopub.execute_input":"2022-01-10T17:39:28.702544Z","iopub.status.idle":"2022-01-10T17:39:28.714057Z","shell.execute_reply.started":"2022-01-10T17:39:28.702477Z","shell.execute_reply":"2022-01-10T17:39:28.713113Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_records, valid_records = parser.parse()","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:28.715788Z","iopub.execute_input":"2022-01-10T17:39:28.716463Z","iopub.status.idle":"2022-01-10T17:39:35.490182Z","shell.execute_reply.started":"2022-01-10T17:39:28.716421Z","shell.execute_reply":"2022-01-10T17:39:35.489420Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"parser.class_map","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:35.491586Z","iopub.execute_input":"2022-01-10T17:39:35.491865Z","iopub.status.idle":"2022-01-10T17:39:35.497873Z","shell.execute_reply.started":"2022-01-10T17:39:35.491815Z","shell.execute_reply":"2022-01-10T17:39:35.496975Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_records[10]","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:35.500660Z","iopub.execute_input":"2022-01-10T17:39:35.500930Z","iopub.status.idle":"2022-01-10T17:39:35.507959Z","shell.execute_reply.started":"2022-01-10T17:39:35.500860Z","shell.execute_reply":"2022-01-10T17:39:35.507158Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"show_record(train_records[1], display_label=True, figsize=(14, 10))","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:35.509645Z","iopub.execute_input":"2022-01-10T17:39:35.509937Z","iopub.status.idle":"2022-01-10T17:39:36.092403Z","shell.execute_reply.started":"2022-01-10T17:39:35.509899Z","shell.execute_reply":"2022-01-10T17:39:36.091757Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## augmentation","metadata":{}},{"cell_type":"code","source":"# Transforms\n# size is set to 384 because EfficientDet requires its inputs to be divisible by 128\nimage_size = 384\ntrain_tfms = tfms.A.Adapter([*tfms.A.aug_tfms(size=image_size, presize=512), tfms.A.Normalize()])\nvalid_tfms = tfms.A.Adapter([*tfms.A.resize_and_pad(image_size), tfms.A.Normalize()])","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:36.094874Z","iopub.execute_input":"2022-01-10T17:39:36.095576Z","iopub.status.idle":"2022-01-10T17:39:36.101622Z","shell.execute_reply.started":"2022-01-10T17:39:36.095539Z","shell.execute_reply":"2022-01-10T17:39:36.100774Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_ds = Dataset(train_records, train_tfms)\nvalid_ds = Dataset(valid_records, valid_tfms)","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:36.103069Z","iopub.execute_input":"2022-01-10T17:39:36.103422Z","iopub.status.idle":"2022-01-10T17:39:36.113571Z","shell.execute_reply.started":"2022-01-10T17:39:36.103378Z","shell.execute_reply":"2022-01-10T17:39:36.111971Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"samples = [train_ds[0] for _ in range(3)]\nshow_samples(samples, ncols=3)","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:36.114773Z","iopub.execute_input":"2022-01-10T17:39:36.116147Z","iopub.status.idle":"2022-01-10T17:39:38.129959Z","shell.execute_reply.started":"2022-01-10T17:39:36.116117Z","shell.execute_reply":"2022-01-10T17:39:38.129169Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 🏗️ creating your model","metadata":{}},{"cell_type":"code","source":"extra_args = {}\nmodel_type = models.ross.efficientdet\nbackbone = model_type.backbones.tf_lite0\n# The efficientdet model requires an img_size parameter\nextra_args['img_size'] = image_size","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:38.131231Z","iopub.execute_input":"2022-01-10T17:39:38.132985Z","iopub.status.idle":"2022-01-10T17:39:38.138089Z","shell.execute_reply.started":"2022-01-10T17:39:38.132945Z","shell.execute_reply":"2022-01-10T17:39:38.136954Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model = model_type.model(backbone=backbone(pretrained=True), num_classes=len(parser.class_map), **extra_args) ","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:38.139720Z","iopub.execute_input":"2022-01-10T17:39:38.139998Z","iopub.status.idle":"2022-01-10T17:39:39.768139Z","shell.execute_reply.started":"2022-01-10T17:39:38.139963Z","shell.execute_reply":"2022-01-10T17:39:39.767393Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_dl = model_type.train_dl(train_ds, batch_size=32, shuffle=True)\nvalid_dl = model_type.valid_dl(valid_ds, batch_size=32, shuffle=False)","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:39.769571Z","iopub.execute_input":"2022-01-10T17:39:39.769840Z","iopub.status.idle":"2022-01-10T17:39:39.774189Z","shell.execute_reply.started":"2022-01-10T17:39:39.769794Z","shell.execute_reply":"2022-01-10T17:39:39.773508Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 👨‍🏫 model training","metadata":{}},{"cell_type":"code","source":"learn = model_type.fastai.learner(dls=[train_dl,valid_dl], model=model)","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:39.775446Z","iopub.execute_input":"2022-01-10T17:39:39.775902Z","iopub.status.idle":"2022-01-10T17:39:39.833795Z","shell.execute_reply.started":"2022-01-10T17:39:39.775865Z","shell.execute_reply":"2022-01-10T17:39:39.833071Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"learn.lr_find()","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:39:39.835300Z","iopub.execute_input":"2022-01-10T17:39:39.835777Z","iopub.status.idle":"2022-01-10T17:43:18.546423Z","shell.execute_reply.started":"2022-01-10T17:39:39.835737Z","shell.execute_reply":"2022-01-10T17:43:18.545720Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"learn.fine_tune(10, 1e-2, freeze_epochs=1)","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:58:47.241410Z","iopub.execute_input":"2022-01-10T17:58:47.241678Z","iopub.status.idle":"2022-01-10T18:41:59.064879Z","shell.execute_reply.started":"2022-01-10T17:58:47.241647Z","shell.execute_reply":"2022-01-10T18:41:59.064175Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# ✨ model inference ","metadata":{"execution":{"iopub.status.busy":"2022-01-10T12:49:56.505812Z","iopub.execute_input":"2022-01-10T12:49:56.506116Z","iopub.status.idle":"2022-01-10T12:49:58.194861Z","shell.execute_reply.started":"2022-01-10T12:49:56.506078Z","shell.execute_reply":"2022-01-10T12:49:58.19423Z"}}},{"cell_type":"code","source":"infer_dl = model_type.infer_dl(valid_ds, batch_size=4, shuffle=False)\npreds = model_type.predict_from_dl(model, infer_dl, keep_images=True,detection_threshold=0.2)","metadata":{"execution":{"iopub.status.busy":"2022-01-10T18:43:11.111930Z","iopub.execute_input":"2022-01-10T18:43:11.112503Z","iopub.status.idle":"2022-01-10T18:43:56.570484Z","shell.execute_reply.started":"2022-01-10T18:43:11.112439Z","shell.execute_reply":"2022-01-10T18:43:56.569799Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"img_number = 10\npred_bboxes_list, gt_bboxes_list = [], []\npred_bboxes = preds[img_number].pred.as_dict()['detection']['bboxes']\nground_truth_bboxes = preds[img_number].ground_truth.as_dict()['detection']['bboxes']\npreds_df = pd.DataFrame(columns=['x','y','width','height'])\ngt_df = pd.DataFrame(columns=['x','y','width','height'])\n\nif len(pred_bboxes) > 0:\n    for box in pred_bboxes:\n        pred_bboxes_list.append([box.xmin,box.ymin,box.xmax - box.xmin,box.ymax - box.ymin])\n    preds_df[['x','y','width','height']] = pred_bboxes_list\n    preds_df = preds_df.astype(int)\n    \nif len(ground_truth_bboxes) > 0:\n    for box in ground_truth_bboxes:\n        gt_bboxes_list.append([box.xmin,box.ymin,box.xmax - box.xmin,box.ymax - box.ymin])\n    gt_df[['x','y','width','height']] = gt_bboxes_list\n    gt_df = gt_df.astype(int)    ","metadata":{"execution":{"iopub.status.busy":"2022-01-10T18:46:06.191353Z","iopub.execute_input":"2022-01-10T18:46:06.191902Z","iopub.status.idle":"2022-01-10T18:46:06.210085Z","shell.execute_reply.started":"2022-01-10T18:46:06.191861Z","shell.execute_reply":"2022-01-10T18:46:06.209204Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(f'predicted bboxes of image-{img_number}')\npreds_df","metadata":{"execution":{"iopub.status.busy":"2022-01-10T18:46:06.763795Z","iopub.execute_input":"2022-01-10T18:46:06.764195Z","iopub.status.idle":"2022-01-10T18:46:06.773768Z","shell.execute_reply.started":"2022-01-10T18:46:06.764162Z","shell.execute_reply":"2022-01-10T18:46:06.773074Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(f'ground truth bboxes of image-{img_number}')\ngt_df","metadata":{"execution":{"iopub.status.busy":"2022-01-10T18:46:07.961544Z","iopub.execute_input":"2022-01-10T18:46:07.961804Z","iopub.status.idle":"2022-01-10T18:46:07.972050Z","shell.execute_reply.started":"2022-01-10T18:46:07.961775Z","shell.execute_reply":"2022-01-10T18:46:07.971327Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"show_preds(preds=preds[10:11],figsize=(15,15))","metadata":{"execution":{"iopub.status.busy":"2022-01-10T18:46:45.695950Z","iopub.execute_input":"2022-01-10T18:46:45.696210Z","iopub.status.idle":"2022-01-10T18:46:46.191566Z","shell.execute_reply.started":"2022-01-10T18:46:45.696180Z","shell.execute_reply":"2022-01-10T18:46:46.190884Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 🔎 references","metadata":{}},{"cell_type":"markdown","source":"## kaggle kernels\n\n1- https://www.kaggle.com/aninda/icevision/data\n\n2- https://www.kaggle.com/nyanswanaung/cots-coco-stratifiedk-5-folds-4919-imgs\n\n## Ice vision official website\n \nhttps://airctic.com/0.11.0/getting_started_object_detection/","metadata":{"execution":{"iopub.status.busy":"2022-01-10T17:37:34.655105Z","iopub.status.idle":"2022-01-10T17:37:34.655769Z","shell.execute_reply.started":"2022-01-10T17:37:34.655506Z","shell.execute_reply":"2022-01-10T17:37:34.655531Z"}}}]}