[Bug] OpenCV imread fails on Windows with Chinese/Unicode file paths

[x2e23t52]

Title:
[Bug] OpenCV imread fails on Windows with Chinese/Unicode file paths

---

Describe the bug
When processing images or videos located in directories containing Chinese (or other Unicode) characters or special symbols (e.g. 空格, (), []), SafeVision fails to load the file. The processing log shows an OpenCV error:

[ WARN:0@0.411] global loadsave.cpp:275 cv::findDecoder imread_('Q:\共享文件夹\X_025_沖田凛花Rinka\课件_014\课件_164\沖田凜花Rinka NO.146 Police Rinka[47P 1V 531.7M]\1 (11).jpg'): can't open/read file: check file path/integrity
Traceback (most recent call last):
  ...
AttributeError: 'NoneType' object has no attribute 'shape'

---

To Reproduce
Steps to reproduce the behavior:

1. Place an image in a directory with Chinese characters and/or special symbols, e.g.

   Q:\共享文件夹\X_025_沖田凛花Rinka\课件_014\课件_164\沖田凜花Rinka NO.146 Police Rinka[47P 1V 531.7M]\1 (11).jpg
   

2. Load this file into SafeVision.

3. Click Start Processing.

---

Expected behavior
The software should be able to handle Unicode/long paths on Windows and successfully process the file (like PIL.Image.open does). Ideally, SafeVision should automatically normalize or copy files into a safe working directory to bypass OpenCV’s imread limitations.

---

Screenshots / Logs

• Screenshot of the UI and error log (attached).

• Key error:

  AttributeError: 'NoneType' object has no attribute 'shape'
  

---

Environment (please complete the following information):

• OS: Windows 10/11
• SafeVision version: v2.0
• Python/OpenCV backend (if relevant): OpenCV 4.x (from bundled exe)
• Input file system: Network drive Q: (SMB) with Chinese/Unicode characters in path

---

Additional context

• The issue seems to be due to OpenCV’s cv::imread not supporting non-ASCII/long paths on Windows.

• Possible solutions:

  • Use PIL.Image.open or cv::imdecode with np.fromfile to support Unicode paths.
  • Provide an option in Settings to copy input files to a safe temp folder with ASCII names before processing.

• Workaround: moving files into a pure-English path (e.g. C:\SV\in\) allows processing to succeed.