PaddleOCR 识别表情包文字示例详解

pip install paddlepaddle
pip install paddleocr

paddleocr==2.5.0.3
paddlepaddle==2.3.1

pip install --upgrade setuptools==59.8.0

import hashlib
import io
from PIL import Image
from paddleocr import PaddleOCR
def get_gif_pic_path(url):
    '''下载图片，返回图片第一帧路径'''
    response = requests.get(url)
    filename = hashlib.md5(url.encode(encoding='utf-8')).hexdigest()
    img = Image.open(io.BytesIO(response.content))
    current = img.tell()
    img_path = f'bqb/{filename}_{current}.png'
    img.save(img_path)
    return img_path
def ocr_get_gif_text_sec(url) -> str:
    '''识别gif第一帧图片的文字'''
    img_path = get_gif_pic_path(url)
    ocr = PaddleOCR(use_angle_cls=True, lang='ch')
    result = ocr.ocr(img_path, cls=True)
    data = []
    for item in result:
        data.append(item[1][0])
    return ",".join(data)
ocr_get_gif_text_sec('https://xxxxxx')

from paddleocr import PaddleOCR
import numpy as np
import cv2
@app.route('/v1/upload_img', methods=['POST'])
def api_v1_upload_img():
    if 'file' not in request.files:
        return "no file part"
    file = request.files['file']
    if file.filename == '':
        return "no selected file"
    if file:
        file_name = file.filename
        # 获取图片的二进制
        res = upload_image(bytes=file.read(), filename=file_name)
        return res
    else:
        return "no file"
def upload_image(bytes, filename = None, mime_type = None):
    ocr = PaddleOCR(use_angle_cls=True, lang="ch")
    ## 核心步骤
    np_arr = np.frombuffer(bytes, dtype=np.uint8)
    img = cv2.imdecode(np_arr, cv2.IMREAD_COLOR)
    # 进行识别
    result = ocr.ocr(img=img, cls=True)
    ocr_result = []
    for line in result:
        ocr_result.append(line[1][0].strip())
    img_content = ' '.join(ocr_result)
    return img_content   

np_arr = np.frombuffer(bytes, dtype=np.uint8)
img = cv2.imdecode(np_arr, cv2.IMREAD_COLOR)