upload

.gitignore

@@ -0,0 +1,11 @@
+__pycache__/
+.ipynb_checkpoints
+
+*.py[cod]
+*$py.class
+.Python
+env/
+build/
+
+data_draw/
+data_crop/

README.md

@@ -0,0 +1,9 @@
+This is a repo for this detailed blog post on [quantran.xyz/blog/building-an-image-classification-model-from-a-to-z/]. Including Grad-CAM visualization Python module ([code](gradcam.py) and [use cases](gradcam-usecase.ipynb)) and notebooks containing training processes and steps. If you want to trace my process, here is the order (for more details you can visit the blog post):
+1. [crop-fix.ipynb]
+2. [google-image-download.ipynb]
+3. [image-augmentation.ipynb]
+4. [resnet-training-with-upsampling.ipynb]
+5. [final-training-all-images.ipynb]
+6. [live-action-model-check.ipynb]
+
+Tools use: [Fastai library which built on top of Pytorch 1.0](https://github.com/fastai/fastai), Matplotlib, scikit-image, [google-images-download](https://github.com/hardikvasa/google-images-download)

crop-fix.ipynb

@@ -0,0 +1,175 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from PIL import Image\n",
+    "from pathlib import Path\n",
+    "from concurrent.futures import ProcessPoolExecutor\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import os"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "data = Path('data')\n",
+    "data_crop = Path('data_crop')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "['pocahontas',\n",
+       " 'mulan',\n",
+       " 'kiki',\n",
+       " 'howl',\n",
+       " 'hercules',\n",
+       " 'mermaid',\n",
+       " 'models',\n",
+       " 'beauty',\n",
+       " 'castle',\n",
+       " 'mononoke',\n",
+       " 'tarzan']"
+      ]
+     },
+     "execution_count": 4,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "classes=os.listdir(str(data))\n",
+    "classes"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "classes = ['pocahontas',\n",
+    " 'mulan',\n",
+    " 'kiki',\n",
+    " 'howl',\n",
+    " 'hercules',\n",
+    " 'mermaid',\n",
+    " 'beauty',\n",
+    " 'castle',\n",
+    " 'mononoke',\n",
+    " 'tarzan']"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# for c in classes:\n",
+    "#     os.mkdir(data_crop/c)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def crop_class(c):\n",
+    "    print(f'Processing {c}...')\n",
+    "    exc=20\n",
+    "    src_path = data/c\n",
+    "    dest_path = data_crop/c\n",
+    "    img_fnames = os.listdir(str(src_path))\n",
+    "    for img_fname in img_fnames:\n",
+    "        img = Image.open(src_path/img_fname)\n",
+    "        cropped = img.crop((exc, exc, img.size[0]-exc, img.size[1]-exc))\n",
+    "        cropped.save(dest_path/img_fname)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "with ProcessPoolExecutor(max_workers=4) as executor:\n",
+    "    for c,_ in zip(classes,executor.map(crop_class,classes)):\n",
+    "        continue\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "True\n",
+      "True\n",
+      "True\n",
+      "True\n",
+      "True\n",
+      "True\n",
+      "True\n",
+      "True\n",
+      "True\n",
+      "True\n"
+     ]
+    }
+   ],
+   "source": [
+    "#sanity check\n",
+    "for c in classes:\n",
+    "    src_path = data/c\n",
+    "    dest_path = data_crop/c\n",
+    "    org = len(os.listdir(str(src_path)))\n",
+    "    fin = len(os.listdir(str(dest_path)))\n",
+    "    print(org==fin)"
+   ]
+  }
+ ],
+ "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": "3.7.1"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}