Echo/fcb_photo_review
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

调整ocr识别的最大长度为1248，过大的长度也可能导致识别错误

Browse Source
This commit is contained in:
liuyebo
2024-07-25 09:51:52 +08:00
parent 6e7e711b16
commit 162aea0ebb
2 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
photo_mask/__init__.py
View File

@@ -29,4 +29,4 @@ SIMILAR_CHAR = {
} }
# 如果不希望识别出空格，可以设置use_space_char=False。做此项设置一定要测试，2.7.3版本此项设置有bug，会导致识别失败 # 如果不希望识别出空格，可以设置use_space_char=False。做此项设置一定要测试，2.7.3版本此项设置有bug，会导致识别失败
OCR = PaddleOCR(use_angle_cls=False, show_log=False, det_db_box_thresh=0.3, det_limit_side_len=2048) OCR = PaddleOCR(use_angle_cls=False, show_log=False, det_db_box_thresh=0.3, det_limit_side_len=1248)

2
visual_model_test/visual_model_test.py
View File

@@ -63,7 +63,7 @@ def visual_model_test(model_type, test_img, task_path, schema):
img["y_offset"] -= offset_y img["y_offset"] -= offset_y
temp_files_paths.append(temp_file.name) temp_files_paths.append(temp_file.name)
parsed_doc = util.get_ocr_layout(PaddleOCR(det_db_box_thresh=0.3, det_limit_side_len=2048), parsed_doc = util.get_ocr_layout(PaddleOCR(det_db_box_thresh=0.3, det_limit_side_len=1248),
temp_file.name) temp_file.name)
# parsed_doc = doc_parser.parse({"doc": temp_file.name})["layout"] # parsed_doc = doc_parser.parse({"doc": temp_file.name})["layout"]
if img["x_offset"] or img["y_offset"]: if img["x_offset"] or img["y_offset"]: