Train_MTL.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from data_loader.MTL_dataloader import get_train_validation_generator\n",
    "from MTL_model import MTL_model as MTL_model_class\n",
    "from callback import CustomCallBacks\n",
    "from evaluation.evaluate import evaluate\n",
    "from tensorflow.keras.optimizers import Adam"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data loader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "det_csv_path = \"/kaggle/input/rsna-pneumonia-detection-challenge/stage_2_train_labels.csv\"\n",
    "seg_csv_path = \"/kaggle/input/siim-acr-pneumothorax-segmentation-data/train-rle.csv\"\n",
    "det_images_path = \"/kaggle/input/rsna-pneumonia-detection-challenge/stage_2_train_images/\"\n",
    "seg_images_path = \"/kaggle/input/siim-acr-pneumothorax-segmentation-data/dicom-images-train/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_gen,val_gen = get_train_validation_generator(det_csv_path,seg_csv_path , det_images_path, seg_images_path)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Build Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "MTL_model_clss= MTL_model_class()\n",
    "MTL_model = MTL_model_clss.MTL_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#from tensorflow.keras.utils import plot_model\n",
    "#plot_model(MTL_model.MTL_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#MTL_model.MTL_model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "INIT_LR = 1e-4\n",
    "EPOCHS =20\n",
    "model_weights_path = \"model_weights.hdf5\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "opt = Adam(lr=INIT_LR, decay=INIT_LR / EPOCHS)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "lossWeights = [1.0,1.0,1.0]\n",
    "MTL_model_clss.MTL_model.compile(optimizer=opt, loss=mtl_loss,loss_weights=lossWeights,  metrics=[])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "history = MTL_model_clss.MTL_model.fit_generator(train_gen,validation_data=val_gen,  callbacks=CustomCallBacks() )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "MTL_model_clss.MTL_model.save_weights(model_weights_path)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Evaluate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "MTL_model_clss.MTL_model.load_weights(model_weights_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "mAP,dice_coef,f1,per,rec,acc = evaluate(val_gen,MTL_model_clss.MTL_model,MTL_model_clss.MTL_model, MTL_model_clss.detector.anchors )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}