{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "Obtener conjuntos de datos de imágenes\n", "==================================" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "
PRECAUCIÓN 😱: El tema presentado en esta sección está clasificado como avanzado. El entendimiento de este contenido es totalmente opcional.
" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Obtener un conjunto de datos para procesamiento de imágenes puede ser un tarea compleja dependiendo del problema a realizar. Muchas compañias tienen políticas de adquisición de datos específicas para resolver problemas complejos o muy de nicho. Sin embargo, en muchos casos las imágenes pueden ser recolectadas desde la misma web de manera automática. En este ejemplo, veremos como realizar esta automatización utilizando la API del motor de búsqueda Google." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Extraer resultados de búsqueda de Google" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Podemos utilizar la API de búsqueda de Google para extraer resultados del motor de búsqueda. Esto incluye también a las imágenes de Google. Esto lo podremos hacer de forma gratuita siempre y cuando no superemos las 100 imágenes descargadas al día.\n", "\n", "> **¡Importante!** Necesitará crear una cuenta en Google Cloud Platform, o GCP, a pesar de utilizar la versión gratuita de este servicio. Note que para crear una cuenta deberá especificar una tarjeta de crédito. **Tenga especial cuidado en no superar el umbral de 100 solicitudes al día para mantenerse en la capa gratuita**." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Habilitando la API de búsqueda de Google" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Para poder utilziar la API de Google, es necesario disponer de las credenciales que permiten interactuar con la misma. Para ello crearemos un nuevo proyecto desde https://console.developers.google.com\n", "\n", "1. Haga click en `ENABLE APIS AND SERVICES`\n", "2. En el cuadro de búsqueda escriba `custom search API`\n", "\n", "\n", "\n", "3. Haga click en `ENABLE`\n", "4. Dirijase a https://console.developers.google.com/apis/credentials donde se administran todas las credenciales para acceder a las APIs de Google.\n", "5. Genere una nueva API key como se muestra\n", "\n", "\n", "\n", "6. Copie la nueva clave generada y peguela en un block de notas para su posterior uso. Esta será su `API KEY`.\n", "\n", "\n", "\n", "### Creando un motor de búsqueda para utilizar en Python \n", "\n", "1. Dirijase a https://cse.google.com/cse/all, donde Google permite crear \"motores\" de busqueda privados para nuestra utilización personal. Crearemos un motor para utilizar en nuestros scripts.\n", "2. Haga click en `ADD`\n", "3. En `Sites to search` escriba `www.google.com` y haga click en `CREATE`\n", "4. Entre al nuevo item que aparece listado y que acaba de crear bajo el nombre `Google`.\n", "5. Copie el valor del campo `Search engine ID` utilizando el boton `COPY TO CLIPBOARD`\n", "6. Habilite la búsqueda de imágenes tildando la opción `IMAGE SEARCH`\n", "\n", "" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Utilizando la API desde Python" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Utilizaremos la librería `Google-Images-Search` la cual está disponible en `pip`\n", "\n", "```\n", "pip install Google-Images-Search\n", "```" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "!pip install google-images-search" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Primero deberemos crear un objeto de tipo `GoogleImagesSearch` para ejecutar las búsquedas de imágenes que necesitamos. En este punto también necesitaremos los dos secretos que generamos más arrivas siendo estos:\n", "\n", "- La clave de acceso a la API, o `API KEY`.\n", "- El identificador del motor de busqueda que creamos, `Search engine ID`" ] }, { "cell_type": "code", "execution_count": 15, "metadata": {}, "outputs": [], "source": [ "from google_images_search import GoogleImagesSearch" ] }, { "cell_type": "code", "execution_count": 31, "metadata": {}, "outputs": [], "source": [ "api_key = \"AIzaXXXXXXXXXXXXXXXXXXXXXXXXXM1LeI\"\n", "search_engine_id = \"05cXXXXXXXXX159\"\n", "gis = GoogleImagesSearch(api_key, search_engine_id)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Especificamos los parámteros de búsqueda:\n", "\n", "- `q`: Las palabras claves que usariamos para la búsqueda.\n", "- `num`: La cantidad máxima de resultas a retornar.\n", "- `safe`: El tipo de filtro de contenido a utilizar (`high`, `medium`, `off`)\n", "- `fileType`: El tipo de archivo que estamos buscando.\n", "- `imgType`: El tipo de imagen que estamos buscando (`clipart`, `photo`, `lineart`) " ] }, { "cell_type": "code", "execution_count": 32, "metadata": {}, "outputs": [], "source": [ "_search_params = {\n", " 'q' : 'barco carguero',\n", " 'num' : 10,\n", " 'safe' : 'high',\n", " 'fileType': 'jpg|png',\n", " 'imgType' : 'photo',\n", "}" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Ejecutamos la búsqueda y le damos un nombre" ] }, { "cell_type": "code", "execution_count": 33, "metadata": {}, "outputs": [], "source": [ "gis.search(search_params=_search_params,custom_image_name='barco_carguero')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Creamos un directorio para guardar los resultados. En general, utilizaremos la siguiente convención: `[nombre_del_dataset]\\[clase]`. En este caso lo llamaremos `barcos\\carguero`. Note que estamos suponiendo que trabajamos en un conjunto de datos llamado \"barcos\", el cual ataca un problema de clasificación de barcos donde una de las posibles clases es \"carguero\"." ] }, { "cell_type": "code", "execution_count": 23, "metadata": {}, "outputs": [], "source": [ "!mkdir -p datasets/barcos/carguero" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Descargamos las imágenes en el directorio indicado:" ] }, { "cell_type": "code", "execution_count": 34, "metadata": {}, "outputs": [], "source": [ "for image in gis.results():\n", " image.download('datasets/barcos/carguero')\n", " image.resize(500, 500)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "> Note que podemos además cambiar el tamaño de las imagenes a dimensiones específicas. Esto puede resultar útil si no queremos hacer demasiado preprocesamiento en nuestros modelos." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Verifiquemos las imágenes descargadas:" ] }, { "cell_type": "code", "execution_count": 35, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\u001b[0m\u001b[01;35m'barco_carguero(1).jpg'\u001b[0m \u001b[01;35m'barco_carguero(5).jpg'\u001b[0m \u001b[01;35m'barco_carguero(9).jpg'\u001b[0m\r\n", "\u001b[01;35m'barco_carguero(2).jpg'\u001b[0m \u001b[01;35m'barco_carguero(6).jpg'\u001b[0m \u001b[01;35mbarco_carguero.jpg\u001b[0m\r\n", "\u001b[01;35m'barco_carguero(3).jpg'\u001b[0m \u001b[01;35m'barco_carguero(7).jpg'\u001b[0m\r\n", "\u001b[01;35m'barco_carguero(4).jpg'\u001b[0m \u001b[01;35m'barco_carguero(8).jpg'\u001b[0m\r\n" ] } ], "source": [ "ls datasets/barcos/carguero" ] }, { "cell_type": "code", "execution_count": 46, "metadata": {}, "outputs": [ { "data": { "image/jpeg": "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", "text/plain": [ "" ] }, "execution_count": 46, "metadata": {}, "output_type": "execute_result" } ], "source": [ "from IPython import display\n", "display.Image(\"./datasets/barcos/carguero/barco_carguero(2).jpg\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Anotaciones\n", "\n", "Disponer de un conjunto de imágenes no necesariamente implica que dispone de un conjunto de datos de imágenes. En particular, es posible que deba invertir más trabajo en las anotaciones. En los problemas de clasificación deberá separar sus imágenes en diferentes directorios indicando a que clase pertencen. En problemas de detección por ejemplo, deberá generar las atonaciones que indican en que regiones de la imágen se encuentran cada uno de los objetos en los que está interesado. En la siguiente sección veremos como lograrlo." ] } ], "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": "3.8.2" } }, "nbformat": 4, "nbformat_minor": 2 }