报告在 CPU 机器上使用 PaddleOCR CPU包, 2小时后遇到内存泄漏问题

[ghost]

描述：
在使用 PaddleOCR 包2小时后，遇到了内存泄漏问题，这是在 CPU 机器上。尽管在循环中处理图像，但 OCR 的内存使用量持续增加，没有任何内存释放，最终导致内存耗尽。

环境：

PaddleOCR 版本：2.73
Python 版本：3.11.8
操作系统：Window 10

• 安装以下包：
  paddlepaddle==2.6.1 -i https://pypi.tuna.tsinghua.edu.cn/simpleol
  paddleocr==2.73
  hanzidentifier==1.1.0
  pillow==10.3.0
  fastapi[all]==0.110.3

期望行为：

OCR 处理期间内存使用量应保持稳定或逐渐增加，但应在处理每个图像后释放，以防止内存耗尽。

[GreatV]

可以提供一个最小可复现demo吗

[tisoz]

同款环境下一样的问题，在本机无泄漏情况，同版本在服务器上就不停增加直至崩溃

[GreatV]

@tisoz 什么样的服务器环境呀，另外可以提供一个最小可复现demo吗

[tisoz]

@tisoz 什么样的服务器环境呀，另外可以提供一个最小可复现demo吗

我去改写下demo，晚点发过来

[tisoz]

@tisoz 什么样的服务器环境呀，另外可以提供一个最小可复现demo吗

环境信息

annotated-types==0.5.0
anyio==3.7.1
astor==0.8.1
asttokens==2.4.1
async-timeout==4.0.2
attrdict==2.0.1
Babel==2.14.0
backcall==0.2.0
bce-python-sdk==0.9.6
beautifulsoup4==4.12.3
better-exceptions==0.3.3
blinker==1.7.0
bs4==0.0.2
cachetools==5.3.3
certifi==2023.7.22
cffi==1.16.0
charset-normalizer==3.3.2
click==8.1.6
colorama==0.4.6
colored==1.4.4
coloredlogs==15.0.1
contourpy==1.1.1
cryptography==42.0.5
cssselect==1.2.0
cssutils==2.10.2
cycler==0.12.1
Cython==3.0.10
DBUtils==3.1.0
ddddocr==1.4.11
decorator==5.1.1
et-xmlfile==1.1.0
exceptiongroup==1.1.2
executing==2.0.1
fastapi==0.100.1
feapder==1.9.0
fire==0.6.0
Flask==3.0.3
flask-babel==4.0.0
flatbuffers==24.3.25
fonttools==4.51.0
future==1.0.0
h11==0.14.0
httpcore==1.0.5
httpx==0.27.0
humanfriendly==10.0
idna==3.4
imageio==2.34.0
imgaug==0.4.0
importlib_metadata==7.1.0
importlib_resources==6.4.0
influxdb==5.3.1
ipython==8.12.3
itsdangerous==2.1.2
jedi==0.19.1
Jinja2==3.1.3
jmespath==1.0.1
kiwisolver==1.4.5
lazy_loader==0.4
lmdb==1.4.1
loguru==0.7.2
lxml==4.9.3
MarkupSafe==2.1.5
matplotlib==3.7.5
matplotlib-inline==0.1.6
mpmath==1.3.0
msgpack==1.0.8
networkx==3.1
numpy==1.24.4
onnxruntime==1.17.3
opencv-contrib-python==4.6.0.66
opencv-python==4.6.0.66
opencv-python-headless==4.9.0.80
openpyxl==3.1.2
opt-einsum==3.3.0
packaging==24.0
paddleocr==2.7.3
paddlepaddle==2.6.1
pandas==2.0.3
parsel==1.9.1
parso==0.8.4
pdf2docx==0.5.8
pickleshare==0.7.5
pillow==10.3.0
premailer==3.10.0
prompt-toolkit==3.0.43
protobuf==3.20.2
psutil==5.9.8
pure-eval==0.2.2
pyclipper==1.3.0.post5
pycparser==2.22
pycryptodome==3.18.0
pydantic==2.1.1
pydantic_core==2.4.0
Pygments==2.17.2
PyMuPDF==1.20.2
PyMuPDFb==1.24.1
PyMySQL==1.1.0
pyparsing==3.1.2
pyperclip==1.8.2
pyreadline3==3.4.1
python-dateutil==2.9.0.post0
python-docx==1.1.0
pytz==2024.1
PyWavelets==1.4.1
PyYAML==6.0.1
rapidfuzz==3.8.1
rarfile==4.2
redis==3.5.3
requests==2.31.0
scikit-image==0.21.0
scipy==1.10.1
shapely==2.0.3
six==1.16.0
sniffio==1.3.0
soupsieve==2.5
stack-data==0.6.3
starlette==0.27.0
sympy==1.12
termcolor==2.4.0
terminal-layout==3.0.0
tifffile==2023.7.10
tqdm==4.66.2
traitlets==5.14.2
typing_extensions==4.7.1
tzdata==2024.1
urllib3==2.2.1
uvicorn==0.23.2
visualdl==2.5.3
w3lib==2.1.2
wcwidth==0.2.13
Werkzeug==3.0.2
win32-setctime==1.1.0
zipp==3.18.1

paddle_request_test.py

import loguru
import requests

log = loguru.logger

if __name__ == '__main__':
    session = requests.session()
    for i in range(1000000):
        ret = session.get("https://www.dmoe.cc/random.php")
        img = ret.content
        ret = session.post("http://127.0.0.1/ocr_img", data=img, headers={"Content-Type": "image/png"})
        log.info(f"{ret.status_code},{ret.text}")

paddle_webapi.py

import loguru
import psutil
from fastapi import FastAPI, Request
from paddleocr import PaddleOCR

ocr = PaddleOCR(lang="ch",
                use_angle_cls=True,
                ocr_version="PP-OCRv4",
                show_log=False,
                use_gpu=False)
log = loguru.logger

app = FastAPI()
app.ocr_info = {}


@app.post("/ocr_img")
async def ocr_img(request: Request):
    # 使用await获取请求体数据
    data = await request.body()

    process = psutil.Process()
    memory_info = process.memory_info()
    memory_usage_mb = memory_info.rss / (1024 ** 2)
    print(len(data))
    result = ocr.ocr(data)

    return {
        "memory_usage_mb": memory_usage_mb,
        "ocr_result": result
    }


if __name__ == '__main__':
    import uvicorn

    uvicorn.run(app, host="0.0.0.0", port=80)

附带日志

C:\Users\Administrator\Desktop\demo>python paddle_request_test.py
2024-05-22 14:58:21.026 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":504.7578125,"ocr_result":[null]}
2024-05-22 14:58:22.220 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":642.70703125,"ocr_result":[null]}
2024-05-22 14:58:22.991 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":657.98046875,"ocr_result":[null]}
2024-05-22 14:58:24.642 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":658.50390625,"ocr_result":[[[[[1236.0,576.0],[1266.0,576.0],[1266.0,596.0],[1236.0,596.0]],["y",0.5495076775550842]],[[[1268.0,613.0],[1314.0,604.0],[1322.0,640.0],[1276.0,649.0]],["A",0.8723329901695251]]]]}
2024-05-22 14:58:25.844 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":678.0234375,"ocr_result":[[[[[1494.0,760.0],[1549.0,751.0],[1554.0,781.0],[1499.0,790.0]],["411",0.8159841895103455]],[[[1426.0,799.0],[1611.0,759.0],[1624.0,815.0],[1438.0,855.0]],["irno",0.9828563928604126]],[[[1386.0,884.0],[1627.0,808.0],[1643.0,855.0],[1401.0,931.0]],["Long CirnoACT.2",0.8822042942047119]],[[[1307.0,940.0],[1618.0,930.0],[1621.0,1016.0],[1310.0,1025.0]],["東方水精妙",0.7978222966194153]],[[[1448.0,1014.0],[1614.0,1014.0],[1614.0,1040.0],[1448.0,1040.0]],["Border of Cirno",0.9584144353866577]]]]}
2024-05-22 14:58:27.932 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":672.03515625,"ocr_result":[null]}
2024-05-22 14:58:28.750 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":672.484375,"ocr_result":[null]}
2024-05-22 14:58:29.536 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":668.859375,"ocr_result":[null]}
2024-05-22 14:58:30.998 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":668.859375,"ocr_result":[[[[[51.0,995.0],[327.0,995.0],[327.0,1021.0],[51.0,1021.0]],["Mega Dimension Neptune Vul",0.9434823989868164]],[[[51.0,1015.0],[329.0,1015.0],[329.0,1043.0],[51.0,1043.0]],["Rave:techt_JNew:world",0.8671271204948425]],[[[53.0,1039.0],[201.0,1039.0],[201.0,1059.0],[53.0,1059.0]],["ILLUSTRATION BY SWD3E2",0.9549381136894226]]]]}
2024-05-22 14:58:32.208 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":651.1171875,"ocr_result":[null]}
2024-05-22 14:58:33.136 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":660.765625,"ocr_result":[[[[[686.0,141.0],[696.0,141.0],[696.0,151.0],[686.0,151.0]],["R",0.5445988774299622]],[[[786.0,143.0],[800.0,143.0],[800.0,153.0],[786.0,153.0]],["N",0.6315249800682068]]]]}
2024-05-22 14:58:34.157 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":666.1015625,"ocr_result":[null]}
2024-05-22 14:58:37.835 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":669.734375,"ocr_result":[null]}
2024-05-22 14:58:39.143 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":669.1015625,"ocr_result":[null]}
2024-05-22 14:58:40.103 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":670.6328125,"ocr_result":[null]}
2024-05-22 14:58:40.858 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":669.6015625,"ocr_result":[null]}
2024-05-22 14:58:41.959 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":670.90625,"ocr_result":[[[[[1692.0,51.0],[1927.0,40.0],[1942.0,341.0],[1707.0,352.0]],["十",0.773455023765564]],[[[1224.0,813.0],[1307.0,879.0],[1269.0,927.0],[1186.0,861.0]],["@Menu",0.8485442996025085]]]]}
2024-05-22 14:58:43.186 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":672.65625,"ocr_result":[null]}
2024-05-22 14:58:43.940 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":672.6796875,"ocr_result":[null]}
2024-05-22 14:58:44.821 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":672.6796875,"ocr_result":[[[[[100.0,502.0],[420.0,502.0],[420.0,544.0],[100.0,544.0]],["MAGUS",0.8599918484687805]],[[[526.0,502.0],[912.0,502.0],[912.0,544.0],[526.0,544.0]],["MARISA",0.8566349148750305]]]]}
2024-05-22 14:58:46.015 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":677.3984375,"ocr_result":[null]}
2024-05-22 14:58:47.227 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":692.546875,"ocr_result":[null]}
2024-05-22 14:58:47.996 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":692.6171875,"ocr_result":[null]}
2024-05-22 14:58:49.335 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":692.89453125,"ocr_result":[null]}
2024-05-22 14:58:50.074 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":695.30078125,"ocr_result":[null]}
2024-05-22 14:58:50.891 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":695.30859375,"ocr_result":[[[[[571.0,545.0],[695.0,541.0],[702.0,773.0],[578.0,777.0]],["联夜",0.8528777360916138]]]]}
2024-05-22 14:58:51.806 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":696.85546875,"ocr_result":[null]}
2024-05-22 14:58:52.717 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":695.65625,"ocr_result":[[[[[1564.0,1046.0],[1912.0,1046.0],[1912.0,1079.0],[1564.0,1079.0]],["Kodoku no Gununu 2nd Anniversary",0.9814635515213013]]]]}
2024-05-22 14:58:53.567 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":701.3125,"ocr_result":[null]}
2024-05-22 14:58:54.384 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":712.01171875,"ocr_result":[null]}
2024-05-22 14:58:55.249 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":712.01171875,"ocr_result":[[[[[1423.0,121.0],[1594.0,185.0],[1579.0,225.0],[1408.0,160.0]],["winT",0.8581411838531494]],[[[1300.0,155.0],[1747.0,335.0],[1727.0,384.0],[1280.0,204.0]],["midnighT",0.9284902811050415]]]]}
2024-05-22 14:58:56.129 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":721.2578125,"ocr_result":[[[[[464.0,284.0],[576.0,284.0],[576.0,808.0],[464.0,808.0]],["楽園の巫女",0.7368157505989075]],[[[438.0,872.0],[607.0,867.0],[608.0,909.0],[439.0,914.0]],[" 博麗靈夢",0.8217113018035889]]]]}
2024-05-22 14:58:57.700 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":726.2578125,"ocr_result":[[[[[725.0,370.0],[795.0,376.0],[791.0,424.0],[721.0,418.0]],["1:25",0.8122374415397644]],[[[804.0,375.0],[908.0,375.0],[908.0,447.0],[804.0,447.0]],["非",0.9911578297615051]],[[[1072.0,371.0],[1142.0,371.0],[1142.0,429.0],[1072.0,429.0]],["欧",0.9997809529304504]],[[[722.0,412.0],[789.0,420.0],[786.0,447.0],[719.0,440.0]],["aocao",0.9073384404182434]]]]}
2024-05-22 14:58:58.677 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":731.26953125,"ocr_result":[[[[[1575.0,10.0],[1724.0,14.0],[1723.0,42.0],[1574.0,37.0]],["ESSENTIA",0.9843231439590454]]]]}
2024-05-22 14:58:59.667 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":741.109375,"ocr_result":[null]}
2024-05-22 14:59:00.454 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":741.734375,"ocr_result":[null]}
2024-05-22 14:59:01.256 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":741.75,"ocr_result":[null]}
2024-05-22 14:59:02.288 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":742.81640625,"ocr_result":[null]}
2024-05-22 14:59:03.642 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":742.43359375,"ocr_result":[null]}
2024-05-22 14:59:06.848 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":743.86328125,"ocr_result":[[[[[1282.0,418.0],[1727.0,428.0],[1724.0,561.0],[1279.0,551.0]],["iTewi",0.9869147539138794]],[[[1092.0,568.0],[1726.0,566.0],[1726.0,621.0],[1092.0,623.0]],["Project Shrine Maiden",0.9945383667945862]]]]}
2024-05-22 14:59:07.791 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":751.46484375,"ocr_result":[null]}
2024-05-22 14:59:08.703 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":762.52734375,"ocr_result":[null]}
2024-05-22 14:59:09.875 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":761.63671875,"ocr_result":[null]}
2024-05-22 14:59:10.662 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":762.93359375,"ocr_result":[null]}
2024-05-22 14:59:11.761 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":762.93359375,"ocr_result":[[[[[865.0,108.0],[950.0,103.0],[953.0,141.0],[868.0,146.0]],["Rah",0.9758593440055847]],[[[1063.0,102.0],[1193.0,130.0],[1183.0,179.0],[1053.0,152.0]],["Yukari",0.942385196685791]],[[[1642.0,100.0],[1785.0,110.0],[1782.0,160.0],[1639.0,150.0]],["Sakuya",0.8806416392326355]],[[[115.0,167.0],[236.0,158.0],[239.0,194.0],[118.0,203.0]],["MoMizi",0.9713937640190125]],[[[296.0,170.0],[373.0,175.0],[370.0,213.0],[293.0,208.0]],["AYa",0.8028512001037598]],[[[466.0,239.0],[583.0,247.0],[580.0,289.0],[463.0,281.0]],["Nitori",0.9491279125213623]],[[[669.0,234.0],[765.0,239.0],[763.0,275.0],[667.0,269.0]],["chen",0.9556978940963745]],[[[1564.0,296.0],[1596.0,296.0],[1596.0,322.0],[1564.0,322.0]],["SM",0.9817856550216675]]]]}
2024-05-22 14:59:12.563 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":772.51953125,"ocr_result":[null]}
2024-05-22 14:59:13.793 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":773.71875,"ocr_result":[null]}
2024-05-22 14:59:14.611 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":773.23046875,"ocr_result":[null]}
2024-05-22 14:59:15.916 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":773.4921875,"ocr_result":[[[[[142.0,383.0],[412.0,383.0],[412.0,461.0],[142.0,461.0]],["MDE.LOT",0.6986165046691895]],[[[280.0,445.0],[412.0,445.0],[412.0,472.0],[280.0,472.0]],["K.SYO.D+",0.9609846472740173]]]]}
2024-05-22 14:59:16.702 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":774.32421875,"ocr_result":[null]}
2024-05-22 14:59:17.456 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":774.375,"ocr_result":[null]}
2024-05-22 14:59:18.479 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":775.91015625,"ocr_result":[null]}
2024-05-22 14:59:19.751 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":774.375,"ocr_result":[null]}
2024-05-22 14:59:20.679 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":782.1640625,"ocr_result":[null]}
2024-05-22 14:59:21.889 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":781.00390625,"ocr_result":[null]}
2024-05-22 14:59:22.867 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":794.828125,"ocr_result":[null]}
2024-05-22 14:59:23.632 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":793.6328125,"ocr_result":[null]}
2024-05-22 14:59:24.472 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":793.9609375,"ocr_result":[[[[[84.0,395.0],[204.0,395.0],[204.0,516.0],[84.0,516.0]],["米",0.501356303691864]]]]}
2024-05-22 14:59:25.692 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":795.328125,"ocr_result":[null]}
2024-05-22 14:59:27.837 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":795.8203125,"ocr_result":[null]}
2024-05-22 14:59:31.169 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":795.78125,"ocr_result":[null]}
2024-05-22 14:59:31.955 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":795.78125,"ocr_result":[[[[[459.0,436.0],[822.0,443.0],[821.0,505.0],[458.0,498.0]],["蓬莱山輝夜",0.9341721534729004]]]]}
2024-05-22 14:59:32.914 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":795.78125,"ocr_result":[[[[[0.0,456.0],[167.0,340.0],[212.0,409.0],[32.0,525.0]],["面",0.7165175080299377]],[[[184.0,478.0],[280.0,478.0],[280.0,556.0],[184.0,556.0]],["食",0.6076569557189941]]]]}
2024-05-22 14:59:33.748 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":802.7109375,"ocr_result":[null]}
2024-05-22 14:59:37.079 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":802.75,"ocr_result":[null]}
2024-05-22 14:59:38.309 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":803.08984375,"ocr_result":[[[[[470.0,735.0],[510.0,735.0],[510.0,766.0],[470.0,766.0]],["Vl",0.61213219165802]],[[[574.0,729.0],[616.0,729.0],[616.0,760.0],[574.0,760.0]],["VI",0.787750244140625]]]]}
2024-05-22 14:59:39.739 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":802.83984375,"ocr_result":[[[[[115.0,317.0],[355.0,331.0],[351.0,410.0],[111.0,397.0]],["星少女",0.9933781027793884]],[[[168.0,405.0],[522.0,405.0],[522.0,445.0],[168.0,445.0]],["自の星空",0.8517228960990906]]]]}
2024-05-22 14:59:40.862 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":810.36328125,"ocr_result":[null]}
2024-05-22 14:59:41.853 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":808.9140625,"ocr_result":[[[[[59.0,145.0],[129.0,132.0],[143.0,202.0],[73.0,216.0]],["?",0.7549221515655518]]]]}
2024-05-22 14:59:44.793 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":807.76171875,"ocr_result":[null]}
2024-05-22 14:59:46.694 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":808.80859375,"ocr_result":[null]}
2024-05-22 14:59:47.810 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":808.08984375,"ocr_result":[[[[[258.0,520.0],[284.0,520.0],[284.0,550.0],[258.0,550.0]],["X",0.5197473168373108]],[[[164.0,727.0],[186.0,727.0],[186.0,746.0],[164.0,746.0]],["?",0.5557878613471985]]]]}
2024-05-22 14:59:48.739 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":805.60546875,"ocr_result":[[[[[79.0,305.0],[149.0,305.0],[149.0,373.0],[79.0,373.0]],["☆",0.9260557889938354]],[[[94.0,390.0],[145.0,390.0],[145.0,430.0],[94.0,430.0]],["☆.",0.7681255340576172]]]]}
2024-05-22 14:59:50.123 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":806.38671875,"ocr_result":[[[[[1402.0,468.0],[1575.0,515.0],[1559.0,571.0],[1387.0,524.0]],["mas",0.7786253094673157]]]]}
2024-05-22 14:59:50.987 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":807.65625,"ocr_result":[[[[[1237.0,122.0],[1463.0,58.0],[1481.0,122.0],[1255.0,186.0]],["花大船王",0.7753020524978638]],[[[1779.0,371.0],[1821.0,368.0],[1837.0,580.0],[1795.0,583.0]],["东港湖料明信员",0.5203613042831421]]]]}
2024-05-22 14:59:52.356 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":807.6328125,"ocr_result":[null]}
2024-05-22 14:59:53.635 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":806.484375,"ocr_result":[null]}
2024-05-22 14:59:54.737 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":820.3515625,"ocr_result":[[[[[1620.0,484.0],[1654.0,484.0],[1654.0,520.0],[1620.0,520.0]],["01",0.9558972120285034]],[[[30.0,1042.0],[60.0,1042.0],[60.0,1056.0],[30.0,1056.0]],["DL",0.6610945463180542]],[[[144.0,1038.0],[222.0,1038.0],[222.0,1060.0],[144.0,1060.0]],["P'or",0.6279547214508057]]]]}
2024-05-22 14:59:55.994 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":825.0859375,"ocr_result":[null]}
2024-05-22 14:59:58.418 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":827.484375,"ocr_result":[[[[[506.0,447.0],[546.0,457.0],[538.0,488.0],[498.0,478.0]],["8",0.839457094669342]]]]}
2024-05-22 14:59:59.377 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":829.16015625,"ocr_result":[[[[[51.0,995.0],[327.0,995.0],[327.0,1021.0],[51.0,1021.0]],["Mega Dimension Neptune Vul",0.9434823989868164]],[[[51.0,1015.0],[329.0,1015.0],[329.0,1043.0],[51.0,1043.0]],["Rave:techt_JNew:world",0.8671271204948425]],[[[53.0,1039.0],[201.0,1039.0],[201.0,1059.0],[53.0,1059.0]],["ILLUSTRATION BY SWD3E2",0.9549381136894226]]]]}
2024-05-22 15:00:00.179 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":837.5546875,"ocr_result":[null]}
2024-05-22 15:00:01.141 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":837.5546875,"ocr_result":[[[[[12.0,276.0],[596.0,280.0],[596.0,322.0],[12.0,318.0]],["NOCHE DE LACRANES",0.888216495513916]]]]}
2024-05-22 15:00:02.009 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":847.390625,"ocr_result":[[[[[828.0,886.0],[888.0,855.0],[938.0,952.0],[878.0,983.0]],["W",0.7900868654251099]]]]}
2024-05-22 15:00:03.034 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":850.69921875,"ocr_result":[null]}
2024-05-22 15:00:04.749 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":850.265625,"ocr_result":[null]}
2024-05-22 15:00:08.068 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":849.1953125,"ocr_result":[null]}
2024-05-22 15:00:08.917 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":983.84375,"ocr_result":[null]}
2024-05-22 15:00:09.687 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":984.359375,"ocr_result":[null]}
2024-05-22 15:00:10.458 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":984.5703125,"ocr_result":[null]}
2024-05-22 15:00:11.733 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":984.58203125,"ocr_result":[null]}
2024-05-22 15:00:13.177 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":996.4375,"ocr_result":[[[[[180.0,799.0],[302.0,808.0],[297.0,872.0],[176.0,863.0]],["7",0.5437024831771851]],[[[422.0,822.0],[527.0,822.0],[527.0,851.0],[422.0,851.0]],["Wman",0.6472772359848022]],[[[400.0,861.0],[510.0,867.0],[509.0,896.0],[398.0,891.0]],["6R282",0.5733041167259216]],[[[407.0,917.0],[457.0,917.0],[457.0,934.0],[407.0,934.0]],[".7",0.6020278334617615]],[[[418.0,959.0],[516.0,959.0],[516.0,983.0],[418.0,983.0]],["aonba",0.6327023506164551]],[[[531.0,964.0],[561.0,964.0],[561.0,980.0],[531.0,980.0]],["641.",0.5768705606460571]],[[[510.0,1029.0],[546.0,1029.0],[546.0,1044.0],[510.0,1044.0]],["11",0.5439799427986145]]]]}
2024-05-22 15:00:13.963 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1004.5234375,"ocr_result":[null]}
2024-05-22 15:00:14.768 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1006.12109375,"ocr_result":[[[[[1030.0,675.0],[1128.0,675.0],[1128.0,881.0],[1030.0,881.0]],["料府",0.7384895086288452]]]]}
2024-05-22 15:00:15.841 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1009.52734375,"ocr_result":[[[[[459.0,458.0],[531.0,460.0],[523.0,670.0],[451.0,667.0]],["熟美铃",0.6489436626434326]],[[[841.0,808.0],[1081.0,802.0],[1083.0,855.0],[843.0,862.0]],[" 東方紅魔卿",0.821448802947998]],[[[806.0,858.0],[1114.0,860.0],[1114.0,893.0],[806.0,891.0]],[" the Embodiment of Scarlet Devil.",0.9856358170509338]]]]}
2024-05-22 15:00:16.752 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1021.0390625,"ocr_result":[null]}
2024-05-22 15:00:17.711 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1023.3671875,"ocr_result":[null]}
2024-05-22 15:00:18.482 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1022.11328125,"ocr_result":[null]}
2024-05-22 15:00:19.346 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1022.16015625,"ocr_result":[[[[[1578.0,564.0],[1726.0,650.0],[1700.0,694.0],[1552.0,607.0]],["止未机",0.6392879486083984]]]]}
2024-05-22 15:00:20.637 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1025.87109375,"ocr_result":[null]}
2024-05-22 15:00:24.079 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1024.79296875,"ocr_result":[[[[[416.0,498.0],[518.0,498.0],[518.0,631.0],[416.0,631.0]],["公",0.5944934487342834]],[[[773.0,869.0],[947.0,874.0],[945.0,934.0],[771.0,929.0]],["Alice",0.890947163105011]],[[[1013.0,941.0],[1126.0,945.0],[1125.0,973.0],[1012.0,968.0]],["Kirisame",0.9317994117736816]]]]}
2024-05-22 15:00:24.865 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1029.9609375,"ocr_result":[null]}
2024-05-22 15:00:25.668 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1030.50390625,"ocr_result":[null]}
2024-05-22 15:00:26.906 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1030.50390625,"ocr_result":[null]}
2024-05-22 15:00:27.786 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1033.078125,"ocr_result":[null]}
2024-05-22 15:00:28.950 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1034.80859375,"ocr_result":[[[[[68.0,22.0],[300.0,22.0],[300.0,50.0],[68.0,50.0]],["ORDERAR",0.9964967370033264]],[[[1288.0,14.0],[1782.0,12.0],[1782.0,46.0],[1288.0,48.0]],["IS THE RDER A RABBIT",0.9430536031723022]]]]}
2024-05-22 15:00:30.170 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1043.66015625,"ocr_result":[null]}
2024-05-22 15:00:30.944 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1047.33203125,"ocr_result":[null]}
2024-05-22 15:00:31.950 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1049.15234375,"ocr_result":[null]}
2024-05-22 15:00:32.782 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1047.96484375,"ocr_result":[null]}
2024-05-22 15:00:33.680 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1048.6328125,"ocr_result":[[[[[980.0,252.0],[1078.0,252.0],[1078.0,298.0],[980.0,298.0]],["ROCK",0.9676804542541504]]]]}
2024-05-22 15:00:38.429 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1049.30078125,"ocr_result":[null]}
2024-05-22 15:00:39.732 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1049.8203125,"ocr_result":[null]}
2024-05-22 15:00:40.722 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1050.8203125,"ocr_result":[null]}
2024-05-22 15:00:41.967 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1049.8203125,"ocr_result":[null]}
2024-05-22 15:00:43.351 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1059.83203125,"ocr_result":[null]}
2024-05-22 15:00:44.136 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1058.484375,"ocr_result":[null]}
2024-05-22 15:00:45.081 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1058.48046875,"ocr_result":[null]}
2024-05-22 15:00:46.402 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1058.48828125,"ocr_result":[null]}
2024-05-22 15:00:47.204 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1058.52734375,"ocr_result":[null]}
2024-05-22 15:00:48.186 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1058.52734375,"ocr_result":[null]}
2024-05-22 15:00:49.225 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1060.86328125,"ocr_result":[null]}
2024-05-22 15:00:49.998 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1060.8125,"ocr_result":[null]}
2024-05-22 15:00:50.847 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1061.31640625,"ocr_result":[[[[[238.0,276.0],[626.0,276.0],[626.0,576.0],[238.0,576.0]],["M",0.5024936199188232]]]]}
2024-05-22 15:00:51.664 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1062.5234375,"ocr_result":[null]}
2024-05-22 15:00:52.437 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1062.52734375,"ocr_result":[null]}
2024-05-22 15:00:53.569 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1063.66796875,"ocr_result":[[[[[1820.0,0.0],[1917.0,3.0],[1914.0,46.0],[1817.0,41.0]],["CRE",0.9959327578544617]],[[[1045.0,66.0],[1144.0,57.0],[1147.0,85.0],[1048.0,94.0]],["00000",0.7149039506912231]],[[[971.0,419.0],[1040.0,430.0],[1035.0,462.0],[966.0,451.0]],["1日限定",0.8982111811637878]],[[[992.0,451.0],[1068.0,467.0],[1060.0,504.0],[984.0,488.0]],["30食！！",0.8545621037483215]]]]}
2024-05-22 15:00:54.402 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1064.46875,"ocr_result":[null]}
2024-05-22 15:00:55.249 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1074.12890625,"ocr_result":[null]}
2024-05-22 15:00:56.051 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1072.82421875,"ocr_result":[null]}
2024-05-22 15:00:56.853 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1075.3046875,"ocr_result":[null]}
2024-05-22 15:00:57.827 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1075.75390625,"ocr_result":[[[[[1423.0,121.0],[1594.0,185.0],[1579.0,225.0],[1408.0,160.0]],["winT",0.8581411838531494]],[[[1300.0,155.0],[1747.0,335.0],[1727.0,384.0],[1280.0,204.0]],["midnighT",0.9284902811050415]]]]}
2024-05-22 15:00:59.106 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1072.0078125,"ocr_result":[[[[[180.0,799.0],[302.0,808.0],[297.0,872.0],[176.0,863.0]],["7",0.5437024831771851]],[[[422.0,822.0],[527.0,822.0],[527.0,851.0],[422.0,851.0]],["Wman",0.6472772359848022]],[[[400.0,861.0],[510.0,867.0],[509.0,896.0],[398.0,891.0]],["6R282",0.5733041167259216]],[[[407.0,917.0],[457.0,917.0],[457.0,934.0],[407.0,934.0]],[".7",0.6020278334617615]],[[[418.0,959.0],[516.0,959.0],[516.0,983.0],[418.0,983.0]],["aonba",0.6327023506164551]],[[[531.0,964.0],[561.0,964.0],[561.0,980.0],[531.0,980.0]],["641.",0.5768705606460571]],[[[510.0,1029.0],[546.0,1029.0],[546.0,1044.0],[510.0,1044.0]],["11",0.5439799427986145]]]]}
2024-05-22 15:01:00.127 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1074.81640625,"ocr_result":[[[[[125.0,212.0],[197.0,210.0],[205.0,419.0],[133.0,422.0]],["ue",0.5999439358711243]],[[[48.0,687.0],[502.0,691.0],[502.0,747.0],[48.0,742.0]],["Azure&Sands",0.9930488467216492]],[[[126.0,744.0],[500.0,744.0],[500.0,792.0],[126.0,792.0]],["八雲紫の多世界解积",0.964676558971405]],[[[42.0,784.0],[276.0,786.0],[276.0,808.0],[42.0,806.0]],["TohoProjectArrange alubm.",0.9830833673477173]],[[[284.0,790.0],[496.0,790.0],[496.0,824.0],[284.0,824.0]],["80.00.9000",0.625348687171936]]]]}
2024-05-22 15:01:01.182 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1089.58203125,"ocr_result":[[[[[725.0,370.0],[795.0,376.0],[791.0,424.0],[721.0,418.0]],["1:25",0.8122374415397644]],[[[804.0,375.0],[908.0,375.0],[908.0,447.0],[804.0,447.0]],["非",0.9911578297615051]],[[[1072.0,371.0],[1142.0,371.0],[1142.0,429.0],[1072.0,429.0]],["欧",0.9997809529304504]],[[[722.0,412.0],[789.0,420.0],[786.0,447.0],[719.0,440.0]],["aocao",0.9073384404182434]]]]}
2024-05-22 15:01:01.952 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1097.8046875,"ocr_result":[null]}
2024-05-22 15:01:02.854 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1098.3046875,"ocr_result":[[[[[236.0,129.0],[404.0,129.0],[404.0,312.0],[236.0,312.0]],["米",0.7585150003433228]]]]}
2024-05-22 15:01:03.661 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1099.4921875,"ocr_result":[null]}
2024-05-22 15:01:08.471 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1100.4375,"ocr_result":[null]}
2024-05-22 15:01:09.321 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1101.78125,"ocr_result":[[[[[1360.0,637.0],[1624.0,582.0],[1638.0,645.0],[1374.0,701.0]],["Gochiusa",0.9909888505935669]]]]}
2024-05-22 15:01:10.766 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1103.1875,"ocr_result":[null]}
2024-05-22 15:01:11.552 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1107.296875,"ocr_result":[null]}
2024-05-22 15:01:12.326 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1107.734375,"ocr_result":[null]}
2024-05-22 15:01:13.144 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1108.23046875,"ocr_result":[null]}
2024-05-22 15:01:14.103 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1110.3125,"ocr_result":[null]}
2024-05-22 15:01:14.874 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1109.23046875,"ocr_result":[null]}
2024-05-22 15:01:15.866 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1111.03125,"ocr_result":[null]}
2024-05-22 15:01:16.872 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1109.55078125,"ocr_result":[[[[[51.0,995.0],[327.0,995.0],[327.0,1021.0],[51.0,1021.0]],["Mega Dimension Neptune Vul",0.9434823989868164]],[[[51.0,1015.0],[329.0,1015.0],[329.0,1043.0],[51.0,1043.0]],["Rave:techt_JNew:world",0.8671271204948425]],[[[53.0,1039.0],[201.0,1039.0],[201.0,1059.0],[53.0,1059.0]],["ILLUSTRATION BY SWD3E2",0.9549381136894226]]]]}
2024-05-22 15:01:18.099 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1115.24609375,"ocr_result":[null]}
2024-05-22 15:01:19.106 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1116.88671875,"ocr_result":[[[[[1820.0,0.0],[1917.0,3.0],[1914.0,46.0],[1817.0,41.0]],["CRE",0.9959327578544617]],[[[1045.0,66.0],[1144.0,57.0],[1147.0,85.0],[1048.0,94.0]],["00000",0.7149039506912231]],[[[971.0,419.0],[1040.0,430.0],[1035.0,462.0],[966.0,451.0]],["1日限定",0.8982111811637878]],[[[992.0,451.0],[1068.0,467.0],[1060.0,504.0],[984.0,488.0]],["30食！！",0.8545621037483215]]]]}
2024-05-22 15:01:20.780 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1120.90625,"ocr_result":[[[[[236.0,68.0],[278.0,67.0],[282.0,210.0],[240.0,212.0]],["?白十字",0.8496513366699219]],[[[246.0,207.0],[274.0,207.0],[274.0,223.0],[246.0,223.0]],["儿C",0.66560959815979]],[[[1144.0,377.0],[1222.0,363.0],[1230.0,408.0],[1152.0,423.0]],["00277",0.6925936937332153]],[[[1058.0,489.0],[1086.0,489.0],[1086.0,608.0],[1058.0,608.0]],["西本寺",0.9935477375984192]],[[[1090.0,483.0],[1124.0,484.0],[1120.0,614.0],[1086.0,613.0]],["净土真宗",0.9520144462585449]],[[[1322.0,507.0],[1358.0,507.0],[1358.0,523.0],[1322.0,523.0]],["防叉站",0.6394058465957642]],[[[1568.0,517.0],[1602.0,518.0],[1598.0,638.0],[1564.0,637.0]],["竹谷光子",0.7018486261367798]],[[[1053.0,629.0],[1117.0,627.0],[1127.0,881.0],[1063.0,883.0]],["法净寺",0.991776168346405]],[[[1563.0,667.0],[1603.0,677.0],[1594.0,711.0],[1554.0,701.0]],["公明党",0.8132438063621521]],[[[354.0,750.0],[478.0,750.0],[478.0,772.0],[354.0,772.0]],["O70250",0.8349987864494324]],[[[1018.0,784.0],[1038.0,784.0],[1038.0,861.0],[1018.0,861.0]],["茶",0.5351760983467102]]]]}
2024-05-22 15:01:21.629 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1117.0625,"ocr_result":[null]}
2024-05-22 15:01:22.633 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1128.16015625,"ocr_result":[null]}
2024-05-22 15:01:23.533 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1126.109375,"ocr_result":[null]}
2024-05-22 15:01:24.288 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1126.3828125,"ocr_result":[null]}
2024-05-22 15:01:25.200 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1126.59765625,"ocr_result":[[[[[84.0,395.0],[204.0,395.0],[204.0,516.0],[84.0,516.0]],["米",0.501356303691864]]]]}
2024-05-22 15:01:26.144 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1120.87890625,"ocr_result":[[[[[21.0,188.0],[368.0,134.0],[385.0,245.0],[38.0,299.0]],["Merry",0.8602188229560852]],[[[191.0,247.0],[651.0,161.0],[669.0,257.0],[209.0,343.0]],["Clritma",0.8685956001281738]]]]}
2024-05-22 15:01:26.917 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1128.05078125,"ocr_result":[null]}
2024-05-22 15:01:27.860 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1130.296875,"ocr_result":[null]}
2024-05-22 15:01:28.634 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1129.046875,"ocr_result":[[[[[1423.0,291.0],[1481.0,289.0],[1493.0,570.0],[1435.0,572.0]],["日本顽張",0.9652426242828369]]]]}
2024-05-22 15:01:29.441 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1130.8828125,"ocr_result":[null]}
2024-05-22 15:01:30.231 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1131.0390625,"ocr_result":[null]}
2024-05-22 15:01:31.020 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1131.0390625,"ocr_result":[null]}
2024-05-22 15:01:31.847 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1131.0390625,"ocr_result":[[[[[7.0,1032.0],[282.0,1039.0],[281.0,1079.0],[6.0,1074.0]],["Jukina Ra Rutherford",0.7308133840560913]],[[[1796.0,1042.0],[1910.0,1042.0],[1910.0,1076.0],[1796.0,1076.0]],["Melancholy",0.8444303274154663]]]]}
2024-05-22 15:01:32.636 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1131.25,"ocr_result":[null]}
2024-05-22 15:01:33.628 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1132.6328125,"ocr_result":[null]}
2024-05-22 15:01:34.649 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1132.69921875,"ocr_result":[null]}
2024-05-22 15:01:37.556 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1131.3671875,"ocr_result":[[[[[1198.0,52.0],[1250.0,52.0],[1250.0,100.0],[1198.0,100.0]],["大",0.7152677178382874]]]]}
2024-05-22 15:01:38.407 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1135.87109375,"ocr_result":[[[[[21.0,188.0],[368.0,134.0],[385.0,245.0],[38.0,299.0]],["Merry",0.8602188229560852]],[[[191.0,247.0],[651.0,161.0],[669.0,257.0],[209.0,343.0]],["Clritma",0.8685956001281738]]]]}
2024-05-22 15:01:39.417 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1138.29296875,"ocr_result":[null]}
2024-05-22 15:01:40.231 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1136.96484375,"ocr_result":[null]}
2024-05-22 15:01:41.047 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1136.96484375,"ocr_result":[null]}
2024-05-22 15:01:41.927 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1138.015625,"ocr_result":[null]}
2024-05-22 15:01:42.917 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1136.96484375,"ocr_result":[null]}
2024-05-22 15:01:43.927 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1138.4375,"ocr_result":[null]}
2024-05-22 15:01:45.184 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1136.96484375,"ocr_result":[null]}
2024-05-22 15:01:46.401 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1136.96875,"ocr_result":[null]}
2024-05-22 15:01:47.312 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1150.47265625,"ocr_result":[null]}
2024-05-22 15:01:48.708 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1151.4609375,"ocr_result":[[[[[802.0,717.0],[991.0,712.0],[992.0,754.0],[803.0,759.0]],["USAN",0.9708397388458252]],[[[827.0,779.0],[975.0,772.0],[977.0,808.0],[829.0,814.0]],["IRCUSP",0.8948941230773926]]]]}
2024-05-22 15:01:49.698 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1152.4140625,"ocr_result":[[[[[58.0,465.0],[166.0,465.0],[166.0,492.0],[58.0,492.0]],["Castle",0.9936686158180237]],[[[101.0,492.0],[200.0,497.0],[199.0,527.0],[100.0,522.0]],["Bravo",0.9966651201248169]],[[[1390.0,599.0],[1471.0,604.0],[1468.0,642.0],[1387.0,637.0]],["黑锦",0.9507646560668945]],[[[1382.0,639.0],[1478.0,639.0],[1478.0,661.0],[1382.0,661.0]],["kuronisiki",0.9830004572868347]]]]}
2024-05-22 15:01:50.484 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1142.03515625,"ocr_result":[null]}
2024-05-22 15:01:51.434 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1143.87109375,"ocr_result":[null]}
2024-05-22 15:01:52.231 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1142.578125,"ocr_result":[null]}
2024-05-22 15:01:53.190 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1143.87109375,"ocr_result":[[[[[1692.0,51.0],[1927.0,40.0],[1942.0,341.0],[1707.0,352.0]],["十",0.773455023765564]],[[[1224.0,813.0],[1307.0,879.0],[1269.0,927.0],[1186.0,861.0]],["@Menu",0.8485442996025085]]]]}
2024-05-22 15:01:53.975 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1144.5703125,"ocr_result":[null]}
2024-05-22 15:01:54.793 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1144.5703125,"ocr_result":[null]}
2024-05-22 15:01:55.594 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1144.6796875,"ocr_result":[null]}
2024-05-22 15:01:56.399 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1144.7109375,"ocr_result":[null]}
2024-05-22 15:01:57.202 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1144.734375,"ocr_result":[null]}
2024-05-22 15:01:57.944 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1144.734375,"ocr_result":[null]}
2024-05-22 15:01:58.921 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1146.203125,"ocr_result":[[[[[98.0,1033.0],[254.0,1033.0],[254.0,1063.0],[98.0,1063.0]],["sler a rablit?",0.7364205121994019]],[[[68.0,1072.0],[134.0,1072.0],[134.0,1095.0],[68.0,1095.0]],["DHATA",0.7088372707366943]]]]}
2024-05-22 15:01:59.753 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1148.23828125,"ocr_result":[null]}
2024-05-22 15:02:00.748 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1159.15625,"ocr_result":[[[[[1402.0,468.0],[1575.0,515.0],[1559.0,571.0],[1387.0,524.0]],["mas",0.7786253094673157]]]]}
2024-05-22 15:02:01.596 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1167.0546875,"ocr_result":[null]}
2024-05-22 15:02:02.857 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1166.98046875,"ocr_result":[null]}
2024-05-22 15:02:03.722 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1165.8046875,"ocr_result":[null]}
2024-05-22 15:02:07.152 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1167.3671875,"ocr_result":[[[[[141.0,883.0],[198.0,883.0],[198.0,940.0],[141.0,940.0]],["★",0.765034019947052]]]]}
2024-05-22 15:02:08.367 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1165.71484375,"ocr_result":[null]}
2024-05-22 15:02:09.156 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1165.984375,"ocr_result":[null]}
2024-05-22 15:02:10.132 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1166.0703125,"ocr_result":[null]}
2024-05-22 15:02:11.500 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1167.63671875,"ocr_result":[null]}
2024-05-22 15:02:12.364 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1166.56640625,"ocr_result":[[[[[686.0,141.0],[696.0,141.0],[696.0,151.0],[686.0,151.0]],["R",0.5445988774299622]],[[[786.0,143.0],[800.0,143.0],[800.0,153.0],[786.0,153.0]],["N",0.6315249800682068]]]]}
2024-05-22 15:02:13.309 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1168.63671875,"ocr_result":[[[[[122.0,544.0],[352.0,544.0],[352.0,584.0],[122.0,584.0]],["milia Scarler",0.8846633434295654]],[[[70.0,577.0],[360.0,584.0],[360.0,616.0],[70.0,609.0]],["TouhouProjecthe Embodim",0.8881730437278748]],[[[394.0,581.0],[516.0,586.0],[516.0,606.0],[394.0,601.0]],["ofScarletDe",0.9711716175079346]]]]}
2024-05-22 15:02:14.205 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1181.84375,"ocr_result":[null]}
2024-05-22 15:02:15.008 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1180.80078125,"ocr_result":[null]}
2024-05-22 15:02:15.872 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1184.97265625,"ocr_result":[null]}
2024-05-22 15:02:16.645 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1183.72265625,"ocr_result":[null]}
2024-05-22 15:02:17.419 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1183.72265625,"ocr_result":[null]}
2024-05-22 15:02:18.451 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1185.11328125,"ocr_result":[null]}
2024-05-22 15:02:19.378 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1184.765625,"ocr_result":[null]}
2024-05-22 15:02:20.558 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1184.83984375,"ocr_result":[[[[[318.0,808.0],[346.0,808.0],[346.0,824.0],[318.0,824.0]],["940",0.5586285591125488]]]]}
2024-05-22 15:02:21.816 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1190.546875,"ocr_result":[null]}
2024-05-22 15:02:22.668 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1191.484375,"ocr_result":[null]}
2024-05-22 15:02:23.433 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1191.484375,"ocr_result":[null]}
2024-05-22 15:02:24.281 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1192.57421875,"ocr_result":[null]}
2024-05-22 15:02:25.238 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1191.484375,"ocr_result":[[[[[865.0,108.0],[950.0,103.0],[953.0,141.0],[868.0,146.0]],["Rah",0.9758593440055847]],[[[1063.0,102.0],[1193.0,130.0],[1183.0,179.0],[1053.0,152.0]],["Yukari",0.942385196685791]],[[[1642.0,100.0],[1785.0,110.0],[1782.0,160.0],[1639.0,150.0]],["Sakuya",0.8806416392326355]],[[[115.0,167.0],[236.0,158.0],[239.0,194.0],[118.0,203.0]],["MoMizi",0.9713937640190125]],[[[296.0,170.0],[373.0,175.0],[370.0,213.0],[293.0,208.0]],["AYa",0.8028512001037598]],[[[466.0,239.0],[583.0,247.0],[580.0,289.0],[463.0,281.0]],["Nitori",0.9491279125213623]],[[[669.0,234.0],[765.0,239.0],[763.0,275.0],[667.0,269.0]],["chen",0.9556978940963745]],[[[1564.0,296.0],[1596.0,296.0],[1596.0,322.0],[1564.0,322.0]],["SM",0.9817856550216675]]]]}
2024-05-22 15:02:26.044 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1192.38671875,"ocr_result":[null]}
2024-05-22 15:02:27.459 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1193.74609375,"ocr_result":[null]}
2024-05-22 15:02:28.351 | INFO     | __main__:<module>:12 - 200,{"memory_usage_mb":1192.38671875,"ocr_result":[null]}

[Ethenlalala]

@tisoz 什么样的服务器环境呀，另外可以提供一个最小可复现demo吗

我去改写下demo，晚点发过来

我在windows上面内存就会一直累加，使用多进程可以解决。

使用memory_profile持续监控，内存变化符合预期。

[GreatV]

@tisoz，感谢提供代码和描述。

[tisoz]

@tisoz 什么样的服务器环境呀，另外可以提供一个最小可复现demo吗

我去改写下demo，晚点发过来

我在windows上面内存就会一直累加，使用多进程可以解决。

使用memory_profile持续监控，内存变化符合预期。

这种实际情况下不太适合内嵌和整合代码，单客户端的时候比较麻烦

[Ethenlalala]

@tisoz 什么样的服务器环境呀，另外可以提供一个最小可复现demo吗

我去改写下demo，晚点发过来

我在windows上面内存就会一直累加，使用多进程可以解决。
使用memory_profile持续监控，内存变化符合预期。

这种实际情况下不太适合内嵌和整合代码，单客户端的时候比较麻烦

是的，因为要重复创建进程，处理速度也会受到影响。只能临时用一下。

[GreatV]

用线程池会不会好一点

[tisoz]

用线程池会不会好一点

2024-05-22 16:22:20.476 | INFO | main::12 - 200,{"memory_usage_mb":5061.09765625,"ocr_result":[null]

截至目前已经增加到了5G，我推测可能是因为导致的问题，但也不好说，因为线程池没有webapi的功能

[GreatV]

设置 FLAGS_allocator_strategy=naive_best_fit 会不会有缓解

PaddleOCR/tools/infer/predict_system.py

Line 22 in e73eb76

os.environ["FLAGS_allocator_strategy"] = "auto_growth"

https://www.paddlepaddle.org.cn/documentation/docs/zh/guides/flags/memory_cn.html

[tisoz]

设置 FLAGS_allocator_strategy=naive_best_fit 会不会有缓解

PaddleOCR/tools/infer/predict_system.py

Line 22 in e73eb76

os.environ["FLAGS_allocator_strategy"] = "auto_growth"

https://www.paddlepaddle.org.cn/documentation/docs/zh/guides/flags/memory_cn.html

没用

[ghost]

可以提供一个最小可复现demo吗

@GreatV

这段最小可复现demo代码的功能是持续监测 /content/img 目录中的图片，并将结果保存在 /content/log 目录中:

import os
import numpy as np
from paddleocr import PaddleOCR, draw_ocr
from PIL import Image
import time

# 设置输入和输出目录
input_dir = '/content/img'
output_dir = '/content/log'

# 初始化 OCR
ocr = PaddleOCR(use_angle_cls=True, lang='en')  # 运行一次以下载并加载模型到内存中

# 进入无限循环以持续进行检测
while True:
    # 遍历输入目录中的图像文件
    for filename in os.listdir(input_dir):
        if filename.endswith('.png') or filename.endswith('.jpg') or filename.endswith('.jpeg'):
            # 读取图像
            img_path = os.path.join(input_dir, filename)
            img = Image.open(img_path).convert('RGB')
            img_np = np.array(img)

            # 进行 OCR
            result = ocr.ocr(img_np, cls=True)

            # 提取 OCR 结果
            res = result[0]
            boxes = [line[0] for line in res]
            txts = [line[1][0] for line in res]
            scores = [line[1][1] for line in res]

            # 在图像上绘制 OCR 结果
            im_show = draw_ocr(img_np, boxes, txts, scores, font_path='/content/Humor-Sans.ttf')
            im_show = Image.fromarray(im_show)

            # 将带有 OCR 结果的图像保存到输出目录中
            output_path = os.path.join(output_dir, filename)
            im_show.save(output_path)

            print(f"检测到 {filename} 中的文本，并将结果保存在 {output_path}")
    
    # 在检查新图像之前暂停一段时间
    time.sleep(1)

[ghost]

运行 36 分钟，从 2.8GB 占用到 4.2GB，并且没有任何释放内存本身的迹象

[GreatV]

我试了一下，把这里修改成下面的，内存应该不会一直增长。

PaddleOCR/tools/infer/predict_system.py

Line 22 in e73eb76

os.environ["FLAGS_allocator_strategy"] = "auto_growth"

os.environ["FLAGS_allocator_strategy"] = "naive_best_fit"
os.environ["FLAGS_eager_delete_tensor_gb"] = "0.0"
os.environ["FLAGS_memory_fraction_of_eager_deletion"] = "1.0"

log: run.log

import loguru
import psutil
from fastapi import FastAPI, Request
from paddleocr import PaddleOCR
from concurrent.futures import ThreadPoolExecutor
import asyncio


ocr = PaddleOCR(lang="ch",
                use_angle_cls=True,
                ocr_version="PP-OCRv4",
                show_log=False,
                use_gpu=False)
log = loguru.logger

app = FastAPI()
app.ocr_info = {}

# 创建一个固定大小的线程池
executor = ThreadPoolExecutor(max_workers=16)

@app.post("/ocr_img")
async def ocr_img(request: Request):
    # 使用await获取请求体数据
    data = await request.body()

    process = psutil.Process()
    memory_info = process.memory_info()
    memory_usage_mb = memory_info.rss / (1024 ** 2)
    # print(len(data))

    # 在线程池中运行ocr.ocr(data)
    loop = asyncio.get_running_loop()
    result = await loop.run_in_executor(executor, ocr.ocr, data)

    return {
        "memory_usage_mb": memory_usage_mb,
        "ocr_result": result
    }


if __name__ == '__main__':
    import uvicorn

    uvicorn.run(app, host="0.0.0.0", port=80)

[GreatV]

另外换一个推理backend，也可以避免这个问题，如onnxruntime、openvino

[Ethenlalala]

我试了一下，把这里修改成下面的，内存应该不会一直增长。

PaddleOCR/tools/infer/predict_system.py

Line 22 in e73eb76

os.environ["FLAGS_allocator_strategy"] = "auto_growth"

os.environ["FLAGS_allocator_strategy"] = "naive_best_fit"
os.environ["FLAGS_eager_delete_tensor_gb"] = "0.0"
os.environ["FLAGS_memory_fraction_of_eager_deletion"] = "1.0"

log: run.log

import loguru
import psutil
from fastapi import FastAPI, Request
from paddleocr import PaddleOCR
from concurrent.futures import ThreadPoolExecutor
import asyncio


ocr = PaddleOCR(lang="ch",
                use_angle_cls=True,
                ocr_version="PP-OCRv4",
                show_log=False,
                use_gpu=False)
log = loguru.logger

app = FastAPI()
app.ocr_info = {}

# 创建一个固定大小的线程池
executor = ThreadPoolExecutor(max_workers=16)

@app.post("/ocr_img")
async def ocr_img(request: Request):
    # 使用await获取请求体数据
    data = await request.body()

    process = psutil.Process()
    memory_info = process.memory_info()
    memory_usage_mb = memory_info.rss / (1024 ** 2)
    # print(len(data))

    # 在线程池中运行ocr.ocr(data)
    loop = asyncio.get_running_loop()
    result = await loop.run_in_executor(executor, ocr.ocr, data)

    return {
        "memory_usage_mb": memory_usage_mb,
        "ocr_result": result
    }


if __name__ == '__main__':
    import uvicorn

    uvicorn.run(app, host="0.0.0.0", port=80)

亲测，还是会持续增加

[GreatV]

我又跑了一会，没有一直增长。

• 一个小时：run.log
• 两个小时：run.log
• 三个小时：run.log
• 四个小时：run.log

[tisoz]

我又跑了一会，没有一直增长。

• 一个小时：run.log
• 两个小时：run.log

我这边还是没效果，会持续增长

[jw-redpanda]

我又跑了一会，没有一直增长。

• 一个小时：run.log
• 两个小时：run.log
• 三个小时：run.log

@GreatV 套用了建议的env，内存没有释放。可以参考一下你的一些环境参数吗？

> systeminfo | findstr /B /C:"OS Name" /C:"OS Version" & python -V & pip freeze | findstr paddle & set | findstr FLAGS_

这是我这边的信息

OS Name:                   Microsoft Windows 11 Pro
OS Version:                10.0.22631 N/A Build 22631
Python 3.11.8
paddleocr==2.7.3
paddlepaddle==2.6.1
FLAGS_allocator_strategy=naive_best_fit
FLAGS_eager_delete_tensor_gb=0.0
FLAGS_memory_fraction_of_eager_deletion=1.0

[GreatV]

****************************************
Paddle version: 2.6.1
Paddle With CUDA: False

OS: macOS 14.5
GCC version: N/A
Clang version: 18.1.5
CMake version: version 3.29.3
Libc version: N/A
Python version: 3.10.14

CUDA version: N/A
cuDNN version: N/A
Nvidia driver version: N/A
Nvidia driver List: N/A
PaddleOCR version: 2.8.0 (in main branch)
****************************************

[GreatV]

@jw-redpanda 修改完predict_sytem.py重新install了嘛

[jw-redpanda]

@jw-redpanda 修改完predict_sytem.py重新install了嘛

@GreatV 我是pip安装的，譬如 pip install paddleocr==2.7.3。
请教一下，怎样安装到最新的paddleocr 2.8.0? 然后重新测试（因为现在最新pypi的是2.7.3）

[GreatV]

@jw-redpanda 把main分支的最新代码拉下来git clone https://github.com/PaddlePaddle/PaddleOCR.git，用python setup.py install 或者 pip install -e . 安装就好了。

[ghost]

@GreatV 是的，使用 pip install paddleocr==2.8.0 后，内存的增加速度变慢了。 非常感谢！

[jw-redpanda]

@jw-redpanda 把main分支的最新代码拉下来git clone https://github.com/PaddlePaddle/PaddleOCR.git，用python setup.py install 或者 pip install -e . 安装就好了。

@GreatV 谢谢！我这边重新测试 paddleocr==2.8.0。大约测试了3个小时，看到内存会先上升然后下降，暂时看不到再在上升到之前的高位。

不过，这里出现了另外一个问题。测试开头的时段，OCR效能大约3秒（每张图片），到了3个小时OCR增加到9秒，之后OCR的效能没有回到之前效能。

一些信息给你参考：
测试时间：大约3小时
测试图片：1445张
OCR效能（开头）：大约3秒
OCR效能（后段）：大约9秒

[MingsYang]

另外换一个推理backend，也可以避免这个问题，如onnxruntime、openvino

v4 导出onnx模型，然后用paddleOCR加载，use_onnx=True, use_gpu=True, 没有使用gpu推理，请问可能有什么原因吗

[SWHL]

这个建议重新建一个Discussion来问哈

[Shubham654]

#11639 (comment)

Facing the same for a long time, tried all the methods but didn't work.

@jw-redpanda @TingquanGao @MingsYang

[Chanli520]

把main分支的最新代码拉下来git clone https://github.com/PaddlePaddle/PaddleOCR.git，用python setup.py install 或者 pip install -e . 安装就好了。 亲测有效！