{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "from os.path import join\n",
    "import os.path as osp\n",
    "import cv2\n",
    "import numpy as np\n",
    "from PIL import Image, ImageDraw\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import math\n",
    "import dlib\n",
    "from pathlib import Path\n",
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "#from utils import imx # Image function extensions\n",
    "#from config import config as cfg\n",
    "\n",
    "import sys\n",
    "sys.path.append('/work/megapixels_dev/megapixels/')\n",
    "from app.settings import app_cfg as cfg\n",
    "from app.utils import im_utils"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.layers import Conv2D, Input, MaxPool2D,Flatten, Dense, Permute, GlobalAveragePooling2D\n",
    "from keras.models import Model\n",
    "from keras.optimizers import adam\n",
    "import numpy as np\n",
    "import pickle\n",
    "import keras\n",
    "import cv2\n",
    "import sys\n",
    "import dlib\n",
    "import os.path\n",
    "from keras.models import Sequential\n",
    "from keras.applications.resnet50 import ResNet50\n",
    "#from keras.applications.resnet50 import Dense\n",
    "from keras.layers import Dense\n",
    "from keras.optimizers import Adam\n",
    "import pickle\n",
    "import numpy as np\n",
    "import cv2\n",
    "import os\n",
    "from keras.layers import Dropout"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model_path = parent_path+\"/common/mmod_human_face_detector.dat\"\n",
    "model_path = '/data_store_hdd/apps/megapixels/models/dlib/mmod_human_face_detector.dat'\n",
    "#cnn_face_detector = dlib.cnn_face_detection_model_v1(model_path)\n",
    "cnn_face_detector = dlib.get_frontal_face_detector()\n",
    "\n",
    "resnet = ResNet50(include_top=False, pooling='avg')\n",
    "model = Sequential()\n",
    "model.add(resnet)\n",
    "model.add(Dense(5, activation='softmax'))\n",
    "model.layers[0].trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fp_model = join(cfg.DIR_MODELS, 'keras', 'model-ldl-resnet.h5')\n",
    "print(fp_model)\n",
    "model.load_weights(fp_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def score_mapping(modelScore):\n",
    "\n",
    "    if modelScore <= 1.9:\n",
    "        mappingScore = ((4 - 2.5) / (1.9 - 1.0)) * (modelScore-1.0) + 2.5\n",
    "    elif modelScore <= 2.8:\n",
    "        mappingScore = ((5.5 - 4) / (2.8 - 1.9)) * (modelScore-1.9) + 4\n",
    "    elif modelScore <= 3.4:\n",
    "        mappingScore = ((6.5 - 5.5) / (3.4 - 2.8)) * (modelScore-2.8) + 5.5\n",
    "    elif modelScore <= 4:\n",
    "        mappingScore = ((8 - 6.5) / (4 - 3.4)) * (modelScore-3.4) + 6.5\n",
    "    elif modelScore < 5:\n",
    "        mappingScore = ((9 - 8) / (5 - 4)) * (modelScore-4) + 8\n",
    "\n",
    "    return mappingScore\n",
    "\n",
    "def beauty_predict(fp_im):\n",
    "    im = cv2.imread(fp_im)\n",
    "\n",
    "    if im.shape[0] > 1280:\n",
    "      new_shape = (1280, im.shape[1] * 1280 / im.shape[0])\n",
    "    elif im.shape[1] > 1280:\n",
    "      new_shape = (im.shape[0] * 1280 / im.shape[1], 1280)\n",
    "    elif im.shape[0] < 640 or im0.shape[1] < 640:\n",
    "      new_shape = (im.shape[0] * 2, im.shape[1] * 2)\n",
    "    else:\n",
    "      new_shape = im.shape[0:2]\n",
    "\n",
    "    im = cv2.resize(im, (int(new_shape[1]), int(new_shape[0])))\n",
    "    dets = cnn_face_detector(im, 0)\n",
    "\n",
    "    for i, d in enumerate(dets):\n",
    "      #face = [d.rect.left(), d.rect.top(), d.rect.right(), d.rect.bottom()]\n",
    "      face = [d.left(), d.top(), d.right(), d.bottom()]\n",
    "      croped_im = im[face[1]:face[3], face[0]:face[2], :]\n",
    "      resized_im = cv2.resize(croped_im, (224, 224))\n",
    "      normed_im = np.array([(resized_im - 127.5) / 127.5])\n",
    "\n",
    "      import time\n",
    "      st = time.time()\n",
    "      pred = model.predict(normed_im)\n",
    "      print(f'{(time.time()-st)}ms')\n",
    "      ldList = pred[0]\n",
    "      out = 1 * ldList[0] + 2 * ldList[1] + 3 * ldList[2] + 4 * ldList[3] + 5 * ldList[4]\n",
    "\n",
    "      out = score_mapping(out)\n",
    "\n",
    "      print(Path(fp_im).stem + \"Score:\" + str('%.2f' % (out)))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "beauty_predict('/home/adam/Downloads/faces/Arnold_Schwarzenegger_0014.jpg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:megapixels]",
   "language": "python",
   "name": "conda-env-megapixels-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}