Imagem CAPTCHA retorna resposta errada: otimização de qualidade

Se CaptchaAI retornar respostas incorretas para CAPTCHAs de imagem, o problema quase sempre está no envio da imagem - e não na solução. Veja como diagnosticar e corrigir isso.

Causas comuns de respostas erradas

Correção 1: verifique a qualidade da imagem antes do envio

import base64
from io import BytesIO
from PIL import Image


def validate_captcha_image(image_path):
    """Check image quality before submitting to CaptchaAI."""
    img = Image.open(image_path)
    width, height = img.size
    issues = []

    # Minimum resolution
    if width < 50 or height < 20:
        issues.append(f"Too small: {width}x{height}px (min 50x20)")

    # Check if mostly blank
    pixels = list(img.getdata())
    if img.mode == "RGB":
        white_count = sum(1 for p in pixels if p[0] > 250 and p[1] > 250 and p[2] > 250)
    else:
        white_count = sum(1 for p in pixels if p > 250)

    blank_ratio = white_count / len(pixels)
    if blank_ratio > 0.95:
        issues.append(f"Image appears blank ({blank_ratio:.0%} white)")

    # File size check
    img_bytes = BytesIO()
    img.save(img_bytes, format="PNG")
    size_kb = img_bytes.tell() / 1024
    if size_kb < 1:
        issues.append(f"File too small ({size_kb:.1f} KB) — may be empty")
    if size_kb > 600:
        issues.append(f"File too large ({size_kb:.0f} KB) — submit under 600 KB")

    return issues


issues = validate_captcha_image("captcha.png")
if issues:
    for issue in issues:
        print(f"WARNING: {issue}")
else:
    print("Image quality OK")

Correção 2: codificação Base64 correta

import base64


def encode_captcha(image_path):
    """Properly encode a CAPTCHA image to base64."""
    with open(image_path, "rb") as f:
        raw = f.read()

    encoded = base64.b64encode(raw).decode("ascii")

    # Verify round-trip
    decoded = base64.b64decode(encoded)
    assert decoded == raw, "Base64 encoding corrupted the image"

    return encoded


# WRONG — encoding a file path string
bad = base64.b64encode(b"captcha.png").decode()  # Encodes filename, not image!

# CORRECT — encoding file contents
with open("captcha.png", "rb") as f:
    good = base64.b64encode(f.read()).decode()

Correção 3: pré-processamento de imagem

from PIL import Image, ImageFilter, ImageEnhance
from io import BytesIO
import base64


def preprocess_captcha(image_path):
    """Improve image quality for better OCR accuracy."""
    img = Image.open(image_path)

    # Convert to RGB if needed
    if img.mode != "RGB":
        img = img.convert("RGB")

    # Upscale small images
    width, height = img.size
    if width < 200:
        scale = 200 / width
        img = img.resize(
            (int(width * scale), int(height * scale)),
            Image.LANCZOS,
        )

    # Increase contrast
    enhancer = ImageEnhance.Contrast(img)
    img = enhancer.enhance(1.5)

    # Sharpen
    img = img.filter(ImageFilter.SHARPEN)

    # Convert to PNG bytes
    buffer = BytesIO()
    img.save(buffer, format="PNG")
    return base64.b64encode(buffer.getvalue()).decode()

Correção 4: adicionar dicas de tipo e idioma

import requests


def solve_image(api_key, image_base64, **hints):
    """Submit image CAPTCHA with quality hints."""
    data = {
        "key": api_key,
        "method": "base64",
        "body": image_base64,
        "json": 1,
    }

    # Add optional hints for better accuracy
    if "language" in hints:
        data["language"] = hints["language"]  # 0=default, 1=Cyrillic, 2=Latin
    if "textinstructions" in hints:
        data["textinstructions"] = hints["textinstructions"]
    if "numeric" in hints:
        data["numeric"] = hints["numeric"]  # 1=digits only, 2=letters only
    if "min_len" in hints:
        data["min_len"] = hints["min_len"]
    if "max_len" in hints:
        data["max_len"] = hints["max_len"]

    resp = requests.post("https://ocr.captchaai.com/in.php", data=data, timeout=30)
    return resp.json()


# Example: Digits-only CAPTCHA, 4-6 characters
result = solve_image(
    "YOUR_API_KEY",
    encoded_image,
    numeric=1,
    min_len=4,
    max_len=6,
)

# Example: Case-sensitive text
result = solve_image(
    "YOUR_API_KEY",
    encoded_image,
    textinstructions="Case-sensitive, enter exactly as shown",
)

Correção 5: capturar elemento CAPTCHA completo

from selenium import webdriver
from selenium.webdriver.common.by import By
import base64


def capture_captcha_element(driver, selector):
    """Screenshot only the CAPTCHA element, not the full page."""
    element = driver.find_element(By.CSS_SELECTOR, selector)

    # Element screenshot (better than page crop)
    png_bytes = element.screenshot_as_png

    # Verify it's not empty
    if len(png_bytes) < 500:
        raise ValueError("Screenshot too small — element may not be visible")

    return base64.b64encode(png_bytes).decode()


# Usage
driver = webdriver.Chrome()
driver.get("https://example.com")
image_b64 = capture_captcha_element(driver, "img#captchaImage")

Correção 6: lidar com CAPTCHAs Dynamic/Rotating

import time


def solve_with_fresh_image(driver, api_key, captcha_selector):
    """Capture and solve CAPTCHA immediately to avoid expiry."""
    # Wait for CAPTCHA to load fully
    time.sleep(2)

    # Capture fresh
    element = driver.find_element(By.CSS_SELECTOR, captcha_selector)
    png_bytes = element.screenshot_as_png
    body = base64.b64encode(png_bytes).decode()

    # Submit immediately
    resp = requests.post("https://ocr.captchaai.com/in.php", data={
        "key": api_key,
        "method": "base64",
        "body": body,
        "json": 1,
    }, timeout=30)
    result = resp.json()

    if result.get("status") != 1:
        raise RuntimeError(result.get("request"))

    task_id = result["request"]

    # Poll — image CAPTCHAs solve fast
    time.sleep(5)
    for _ in range(12):
        resp = requests.get("https://ocr.captchaai.com/res.php", params={
            "key": api_key, "action": "get",
            "id": task_id, "json": 1,
        }, timeout=15)
        data = resp.json()

        if data.get("status") == 1:
            return data["request"]
        if data["request"] != "CAPCHA_NOT_READY":
            raise RuntimeError(data["request"])
        time.sleep(3)

    raise TimeoutError("Image solve timeout")

Lista de verificação de solução de problemas

Perguntas frequentes

Quão preciso é CaptchaAI para CAPTCHAs de imagem?

Com imagens enviadas corretamente, CaptchaAI suporta mais de 27.500 tipos de CAPTCHA com alta precisão. A maioria das falhas se deve à má qualidade da imagem ou a parâmetros incorretos.

Devo pré-processar as imagens antes de enviar?

Somente se a imagem original for de baixa qualidade. CaptchaAI lida bem com imagens CAPTCHA padrão sem pré-processamento. Aumentar a escala de imagens muito pequenas e aumentar o contraste pode ajudar em casos extremos.

Posso relatar respostas erradas?

Sim. Use o endpoint reportbad com o ID da tarefa para relatar respostas incorretas. Isso ajuda a aprimorar a precisão e pode creditar sua conta.

Guias Relacionados

• Diagnóstico de queda na taxa de sucesso
• Validação de solicitação de parâmetros incorretos

• Resolva imagens com precisão -experimente CaptchaAI.*