Grid Image CAPTCHA : mappage de coordonnées et sélection de cellules

Les CAPTCHA d'image de grille - comme les défis d'image reCAPTCHA v2 - présentent une grille 3×3 ou 4×4 et demandent aux utilisateurs de sélectionner les cellules correspondant à une instruction (« Sélectionner tous les carrés avec des feux de circulation »). CaptchaAI renvoie les index des cellules. Ce guide explique comment capturer la grille, mapper les cellules aux coordonnées et cliquer sur les vignettes appropriées.

Dispositions de grille

Les CAPTCHA utilisent deux tailles de grille standard :

3×3 Grid:          4×4 Grid:
1  2  3            1   2   3   4
4  5  6            5   6   7   8
7  8  9            9  10  11  12
                   13  14  15  16

Les cellules sont numérotées de gauche à droite et de haut en bas – ordre de lecture.

Étape 1 : Capturer l'image de la grille

Python (sélénium)

import base64
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.support import expected_conditions as EC

driver = webdriver.Chrome()
driver.get("https://example.com/form")

# Wait for reCAPTCHA iframe
WebDriverWait(driver, 10).until(
    EC.presence_of_element_located((By.CSS_SELECTOR, "iframe[src*='recaptcha']"))
)

# Switch to challenge iframe
iframes = driver.find_elements(By.CSS_SELECTOR, "iframe[src*='recaptcha']")
challenge_iframe = iframes[-1]  # Challenge iframe is typically the last one
driver.switch_to.frame(challenge_iframe)

# Get the grid image
grid_img = driver.find_element(By.CSS_SELECTOR, "img.rc-image-tile-33, img.rc-image-tile-44")
img_src = grid_img.get_attribute("src")

# Get instruction text
instruction = driver.find_element(
    By.CSS_SELECTOR, ".rc-imageselect-desc-wrapper"
).text
print(f"Instruction: {instruction}")

# Screenshot the grid as base64
img_b64 = grid_img.screenshot_as_base64

# Determine grid size
classes = grid_img.get_attribute("class")
grid_size = "4x4" if "44" in classes else "3x3"
print(f"Grid size: {grid_size}")

driver.switch_to.default_content()

JavaScript (Marionnette)

const puppeteer = require('puppeteer');
const fs = require('fs');

const browser = await puppeteer.launch({ headless: false });
const page = await browser.newPage();
await page.goto('https://example.com/form');

// Find the challenge iframe
const frames = page.frames();
const challengeFrame = frames.find(f => f.url().includes('recaptcha'));

// Get instruction
const instruction = await challengeFrame.$eval(
  '.rc-imageselect-desc-wrapper',
  el => el.textContent.trim()
);

// Screenshot the grid image
const gridImg = await challengeFrame.$('img.rc-image-tile-33, img.rc-image-tile-44');
const imgBuffer = await gridImg.screenshot();
const imgBase64 = imgBuffer.toString('base64');

// Determine grid size
const className = await challengeFrame.$eval(
  'img.rc-image-tile-33, img.rc-image-tile-44',
  el => el.className
);
const gridSize = className.includes('44') ? '4x4' : '3x3';
console.log(`Grid: ${gridSize}, Instruction: ${instruction}`);

Étape 2 : Soumettre à CaptchaAI

import requests
import time
import json

API_KEY = "YOUR_API_KEY"

# Parse the instruction to a simple keyword
# "Select all images with traffic lights" → "traffic lights"
import re
keyword_match = re.search(r'(?:with|of|containing)\s+(.+?)\.?$', instruction, re.I)
keyword = keyword_match.group(1) if keyword_match else instruction

# Submit
with open("/tmp/grid.png", "wb") as f:
    f.write(base64.b64decode(img_b64))

with open("/tmp/grid.png", "rb") as f:
    resp = requests.post("https://ocr.captchaai.com/in.php", 
        files={"file": f},
        data={
            "key": API_KEY,
            "method": "post",
            "grid_size": grid_size,
            "img_type": "recaptcha",
            "instructions": keyword,
            "json": "1",
        }
    ).json()

if resp["status"] != 1:
    raise Exception(f"Submit error: {resp['request']}")

task_id = resp["request"]

# Poll
for _ in range(20):
    time.sleep(5)
    result = requests.get("https://ocr.captchaai.com/res.php", params={
        "key": API_KEY, "action": "get", "id": task_id, "json": "1"
    }).json()

    if result["status"] == 1:
        cells = json.loads(result["request"])
        print(f"Cells to click: {cells}")  # e.g., [1, 3, 6, 9]
        break
    if result["request"] != "CAPCHA_NOT_READY":
        raise Exception(f"Error: {result['request']}")

Étape 3 : Mappez les indices de cellules pour cliquer sur les coordonnées

Convertissez les indices de cellule de base 1 en coordonnées de pixels dans la grille :

def cell_to_coordinates(cell_index, grid_size, grid_width, grid_height):
    """Convert a 1-based cell index to (x, y) center coordinates."""
    if grid_size == "3x3":
        cols, rows = 3, 3
    else:
        cols, rows = 4, 4

    cell_w = grid_width / cols
    cell_h = grid_height / rows

    # Convert 1-based index to 0-based row/col
    idx = cell_index - 1
    col = idx % cols
    row = idx // cols

    # Center of the cell
    x = col * cell_w + cell_w / 2
    y = row * cell_h + cell_h / 2

    return int(x), int(y)

# Example: grid is 300×300
for cell in cells:
    x, y = cell_to_coordinates(cell, grid_size, 300, 300)
    print(f"Cell {cell} → ({x}, {y})")

Sortie pour une grille 3×3 (300×300) :

Cell 1 → (50, 50)
Cell 3 → (250, 50)
Cell 6 → (250, 150)
Cell 9 → (250, 250)

Étape 4 : Cliquez sur les cellules

Sélénium

from selenium.webdriver.common.action_chains import ActionChains

driver.switch_to.frame(challenge_iframe)

# Get grid element position and size
grid_el = driver.find_element(By.CSS_SELECTOR, ".rc-imageselect-target")
grid_rect = grid_el.rect
grid_w = grid_rect["width"]
grid_h = grid_rect["height"]

actions = ActionChains(driver)

for cell in cells:
    x, y = cell_to_coordinates(cell, grid_size, grid_w, grid_h)
    # Click relative to grid element's top-left corner
    actions.move_to_element_with_offset(
        grid_el,
        x - grid_w / 2,  # offset from center
        y - grid_h / 2
    ).click()

actions.perform()

# Click verify
verify_btn = driver.find_element(By.ID, "recaptcha-verify-button")
verify_btn.click()

driver.switch_to.default_content()

Marionnettiste

// Click each cell by index
const tableRows = await challengeFrame.$$('table.rc-imageselect-table tr');
for (const cellIdx of cells) {
  const row = Math.floor((cellIdx - 1) / (gridSize === '4x4' ? 4 : 3));
  const col = (cellIdx - 1) % (gridSize === '4x4' ? 4 : 3);
  const cell = (await tableRows[row].$$('td'))[col];
  await cell.click();
  await new Promise(r => setTimeout(r, 200));
}

await challengeFrame.click('#recaptcha-verify-button');

Gestion des tuiles dynamiques

Certaines grilles reCAPTCHA v2 remplacent les vignettes cliquées par de nouvelles images. Gérez cela avec une boucle de nouvelle tentative :

def solve_with_dynamic_tiles(driver, api_key, max_rounds=3):
    for round_num in range(max_rounds):
        driver.switch_to.frame(challenge_iframe)

        # Re-capture grid and instruction
        img_b64 = driver.find_element(
            By.CSS_SELECTOR, "img.rc-image-tile-33"
        ).screenshot_as_base64

        # Submit and get cells (same as above)
        cells = submit_and_poll(api_key, img_b64, "3x3", keyword)

        if not cells:
            break

        # Click cells
        click_cells(driver, cells, "3x3")

        # Click verify
        driver.find_element(By.ID, "recaptcha-verify-button").click()

        driver.switch_to.default_content()
        time.sleep(2)

        # Check if solved (no more challenge iframe)
        try:
            driver.switch_to.frame(challenge_iframe)
            driver.switch_to.default_content()
        except Exception:
            return True  # Solved

    return False

Dépannage

FAQ

CaptchaAI prend-il en charge les grilles 4×4 ?

Oui. Définissez grid_size=4x4 et la réponse utilisera les indices 1 à 16.

Quelle est la précision de la résolution d’images en grille ?

La précision dépend de la qualité de l'image. Envoyez l’image CAPTCHA originale sans recadrage ni compression.

Résolvez les CAPTCHA d’image de grille avec CaptchaAI

Obtenez votre clé API surcaptchaai.com.

Guides associés

• Code d'instructions BLS CAPTCHA
• Extraction des paramètres reCAPTCHA
• Prétraitement des images CAPTCHA