[Fix] Update image decoding and processing in Worker class

[GreatV]

This pull request includes updates to improve the functionality and accuracy of OCR processing in the PPOCRLabel and autoDialog modules. The key changes involve modifying configuration parameters and fixing how images are handled during OCR prediction.

Changes to OCR configuration:

• In PPOCRLabel.py, the get_str method now sets the use_textline_orientation parameter to False, disabling textline orientation detection. This change likely reflects a shift in how text orientation is handled in the OCR pipeline.

Fixes to image handling:

• In libs/autoDialog.py, the run method now ensures that images are properly decoded before shape extraction and OCR prediction. The cv2.imdecode function's output is stored in a variable (img) before accessing its shape, and the decoded image itself is passed to the OCR prediction method instead of the file path. This improves the robustness of the image processing workflow.

[Copilot]

Pull Request Overview

This PR refines OCR image handling in libs/autoDialog.py for robustness and adjusts OCR orientation settings in PPOCRLabel.py.

• Disable textline orientation detection in OCR configuration to streamline orientation handling.
• Decode images into memory before extracting dimensions and passing them to OCR, improving error handling.

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
libs/autoDialog.py	Decode image to img variable, extract shape, and pass img to OCR instead of file path
PPOCRLabel.py	Set use_textline_orientation to False in OCR parameters

Comments suppressed due to low confidence (1)

libs/autoDialog.py:48

• [nitpick] Consider adding unit tests for both branches of this size check (images smaller and larger than 32×32) to verify correct handling of edge cases.

                        if h > 32 and w > 32:

[Copilot]

cv2.imdecode can return None when decoding fails; add a check for img is None before accessing img.shape to avoid runtime errors.

Suggested change

	)
	)
	if img is None:
	logger.warning("Failed to decode image: %s", img_path)
	self.result_dic = None
	continue