{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Automatic photo categorization\n",
    "\n",
    "\n",
    "Goals:\n",
    "  - Categorize photos into semantically similar groups.\n",
    "  - Mark similar photos for removal.\n",
    "\n",
    "\n",
    "## Table of contents\n",
    " 1. [Features](#features)\n",
    " 2. [Clustering](#clustering)\n",
    " 3. [Deduplication](#deduplication)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "from tqdm import tqdm\n",
    "#from tqdm.notebook import tqdm\n",
    "\n",
    "from toolz import compose\n",
    "from toolz.curried import map, filter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from photocat import fs, photo, group\n",
    "\n",
    "INPUT_DIR = 'data/photos'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a name=\"features\"></a>\n",
    "## Features\n",
    "\n",
    "Extract features from EXIF data and YOLOv3 output."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def show_photos(photos, n_row, n_col, size=4):\n",
    "    _, axs = plt.subplots(n_row, n_col, figsize=(n_col*size, n_row*size))\n",
    "    axs = axs.flatten()\n",
    "    for p, ax in zip(photos, axs):\n",
    "        ax.imshow(p.thumbnail)\n",
    "    plt.show()\n",
    "\n",
    "photos = compose(\n",
    "    list,\n",
    "    tqdm,\n",
    "    map(lambda f: photo.Photo(f)),\n",
    "    fs.list_images\n",
    ")(INPUT_DIR)\n",
    "\n",
    "show_photos(photos[0:24], 6, 4)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a name=\"clustering\"></a>\n",
    "## Clustering\n",
    "\n",
    "Normalize features and cluster with DBSCAN."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a name=\"deduplication\"></a>\n",
    "## Deduplication\n",
    "\n",
    "Use eucledian distance between outputs of topmost YOLOv3 layers as a metric for photo similarity."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}