Process finished with exit code -1073741795 (0xC000001D)

[Michaelufcb]

Description of the problem:
EasyOcr is successfully installed, the program starts, but the text from the picture does not output at all.
Code:
import easy ocr
Reader = easyocr.Reader(['ru', 'en']) # this needs to be run only once to load the model into memory
result = Reader.readtext('12.jpg ')

Conclusion:
neither CUDA nor MPS are available — the CPU is used by default. Note. This module works much faster with a GPU.

The process is completed with exit code -1073741795 (0xC000001D).

My torment:
I ran into a big problem when using easyOCR, namely, when working on a laptop with an i5 -8250U processor, I installed and tested your library for the first time, it started almost immediately without problems and recognized the text with the image, which I was very happy about.
Developing a program for classifying PDF files by keywords) At the end of the practice, I threw the virtual environment with this project on a flash drive, then started running it on an old laptop (i3-2100m, GT-610m) and the library refused to work on it, then I tried to run it on a PC (i7-4960X, RTX 2060, 64 RAM). I spent 10 hours trying to run this library, in the end I didn't succeed, the attempts I made during these 10 hours:

Reinstalling EasyOCR
Reinstalling PIL, V2, Torch
I was poking around in the code, I didn't understand anything
Created a new virtual environment, reinstalled everything, nothing helped
Old Python and other Python versions are installed.
Changed dependency versions randomly
I tried to install it very carefully several times according to the manual:
pip install torch torchvision torchaudio
pip install easyocr
And it didn't help, it outputs "The process is completed with exit code -1073741795 (0xC000001D)
I don't know what to do anymore, but I'm asking for help with this problem.

[Michaelufcb]

[SIA86]

I have the similar problem. But the code ends with no errors at all.
Win10
Python 3.11.3
easyocr=1.7.0
torch=2.0.1+cpu
cv2=4.8.0

If debugging and put the the line 5 to debug console then got the [WinError 100054]

With step-by-step debugging this function was the last one:

[lvzt]

Hey, I updated easyocr from 1.7.0 to 1.7.1. The problem is still there. I'm not sure the reason is my OS. I'm using Windows 10 Home edition.

[Michaelufcb]

Hello everyone I am writing 4 days later, during this time I have assembled a new system - I switched from i7 4960x to i7 7800X and you know what? Of course, EasyOcr was launched the first time without dancing with a tambourine. In general, the point here is not in the OS and not in the libraries, but directly in the stone. In the wilds of the Internet, I found that this code is associated with the absence of some technologies, instructions in hardware. I killed 25 hours to solve this problem, you can try to solve the problem with some script, I don't know.
P.S. I found out experimentally that even after compilation, the program does not run on old processors, that is, it's definitely not about the interpreter, etc. I also tested it on Intel Core i3 2310M. I hope I will save someone time.

[rxlfnng]

I resolved the problem, I exported a recongnition model to onnx, then it worked.Read this for reference.#746

[rpmrider]

I saw this message too late. I also spent a lot of time on the same issue. It turns out that EasyOCR does not work properly on older CPUs. The CPU I tested on is quite old, but I did not anticipate encountering this problem.

The test code is very simple:
import easyocr
import cv2

image_path = "test_image.png"
print("debug - 1")
reader = easyocr.Reader(['en'], gpu=False)
print("debug - 2")
image = cv2.imread(image_path)
print("debug - 3")
results = reader.readtext(image_path)
print("debug - 4")

The issue is that "debug - 4" is never printed, and there is no error message displayed.

Failed Test CPUs:
Xeon E3-1230 V2: Failed (4/8 cores) Released Q2 2011
ATOM X5-Z8350: Failed (4 cores) Released Q1 2016

Successful Test CPU:
i7-10700K: Success (8/16 cores) Released Q2 2020

Unfortunately, I do not have access to more hardware for further testing. Could anyone test this on the following CPUs?
AMD-5500U, AMD-7840HS, Intel N100