Skip to content

周伟奇 / bmw-ocr

Go to a project
Toggle navigation
Toggle navigation pinning
8cf3f917 by 周伟奇
Options

fix mvc slice

1 parent 57a6afc2
Inline Side-by-side
Showing 2 changed files with 78 additions and 14 deletions
src/apps/doc/management/commands/ocr_process.py
...@@ -452,14 +452,17 @@ class Command(BaseCommand, LoggerMixin): ...@@ -452,14 +452,17 @@ class Command(BaseCommand, LoggerMixin):
452 detail_dict.get('words', '')) 452 detail_dict.get('words', ''))
453 tmp_position_list = detail_dict.get('position', [0, 0, 0, 0]) 453 tmp_position_list = detail_dict.get('position', [0, 0, 0, 0])
454 if len(tmp_position_list) == 4: 454 if len(tmp_position_list) == 4:
455 position_dict[cn_key] = { 455 position_dict.setdefault(cn_key, []).append(
456 consts.FIELD_POSITION_KEY: { 456 {
457 'top': tmp_position_list[1], 457 consts.FIELD_POSITION_KEY: {
458 'left': tmp_position_list[0], 458 'top': tmp_position_list[1],
459 'height': tmp_position_list[-1], 459 'left': tmp_position_list[0],
460 'width': tmp_position_list[2], 460 'height': tmp_position_list[-1],
461 'width': tmp_position_list[2],
462 }
461 } 463 }
462 } 464 )
465
463 rebuild_data_dict[consts.ALL_POSITION_KEY_2] = position_dict 466 rebuild_data_dict[consts.ALL_POSITION_KEY_2] = position_dict
464 rebuild_data_dict[consts.IMG_PATH_KEY_2] = img_path 467 rebuild_data_dict[consts.IMG_PATH_KEY_2] = img_path
465 rebuild_data_dict[consts.SECTION_IMG_PATH_KEY_2] = section_img_path 468 rebuild_data_dict[consts.SECTION_IMG_PATH_KEY_2] = section_img_path
......
src/celery_compare/tasks.py
...@@ -373,8 +373,8 @@ def ca_compare_license(license_en, ocr_res_dict, field_list): ...@@ -373,8 +373,8 @@ def ca_compare_license(license_en, ocr_res_dict, field_list):
373 if ocr_res_str is not None: 373 if ocr_res_str is not None:
374 ocr_res_list = json.loads(ocr_res_str) 374 ocr_res_list = json.loads(ocr_res_str)
375 375
376 # 副页去除 TODO 3/4页去除 376 # 副页去除 3/4页去除
377 if ocr_field == consts.DL_OCR_FIELD: 377 if ocr_field == consts.DL_OCR_FIELD or ocr_field == consts.MVC_OCR_FIELD:
378 tmp_list = [] 378 tmp_list = []
379 for res in ocr_res_list: 379 for res in ocr_res_list:
380 if compare_logic['vinNo'][0] in res: 380 if compare_logic['vinNo'][0] in res:
...@@ -1433,6 +1433,15 @@ def se_compare_license(license_en, ocr_res_dict, field_list): ...@@ -1433,6 +1433,15 @@ def se_compare_license(license_en, ocr_res_dict, field_list):
1433 ocr_res_str = ocr_res_dict.get(ocr_field) 1433 ocr_res_str = ocr_res_dict.get(ocr_field)
1434 if ocr_res_str is not None: 1434 if ocr_res_str is not None:
1435 ocr_res_list = json.loads(ocr_res_str) 1435 ocr_res_list = json.loads(ocr_res_str)
1436
1437 # 3/4页去除
1438 if ocr_field == consts.MVC_OCR_FIELD:
1439 tmp_list = []
1440 for res in ocr_res_list:
1441 if compare_logic['vinNo'][0] in res:
1442 tmp_list.append(res)
1443 ocr_res_list = tmp_list
1444
1436 length = len(ocr_res_list) 1445 length = len(ocr_res_list)
1437 1446
1438 # 身份证、居住证 过期期限特殊处理 1447 # 身份证、居住证 过期期限特殊处理
...@@ -1571,6 +1580,7 @@ def se_mvc34_compare(license_en, ocr_res_dict, field_list): ...@@ -1571,6 +1580,7 @@ def se_mvc34_compare(license_en, ocr_res_dict, field_list):
1571 1580
1572 is_find = False 1581 is_find = False
1573 result_field_list = [] 1582 result_field_list = []
1583 field_img_path_dict = dict()
1574 1584
1575 if ocr_res_str is not None: 1585 if ocr_res_str is not None:
1576 ocr_res_list = json.loads(ocr_res_str) 1586 ocr_res_list = json.loads(ocr_res_str)
...@@ -1580,7 +1590,8 @@ def se_mvc34_compare(license_en, ocr_res_dict, field_list): ...@@ -1580,7 +1590,8 @@ def se_mvc34_compare(license_en, ocr_res_dict, field_list):
1580 first_res = None 1590 first_res = None
1581 for res_idx in range(length-1, -1, -1): 1591 for res_idx in range(length-1, -1, -1):
1582 if consts.TRANSFER_DATE in ocr_res_list[res_idx]: 1592 if consts.TRANSFER_DATE in ocr_res_list[res_idx]:
1583 img_path = ocr_res_list[res_idx].get(consts.IMG_PATH_KEY, '') 1593 img_path = ocr_res_list[res_idx].get(consts.IMG_PATH_KEY_2, '')
1594 section_img_path = ocr_res_list[res_idx].get(consts.SECTION_IMG_PATH_KEY_2, '')
1584 for idx, transfer_date in enumerate(ocr_res_list[res_idx].get(consts.TRANSFER_DATE, [])): 1595 for idx, transfer_date in enumerate(ocr_res_list[res_idx].get(consts.TRANSFER_DATE, [])):
1585 try: 1596 try:
1586 transfer_name = ocr_res_list[res_idx].get(consts.TRANSFER_NAME, [])[idx] 1597 transfer_name = ocr_res_list[res_idx].get(consts.TRANSFER_NAME, [])[idx]
...@@ -1590,11 +1601,32 @@ def se_mvc34_compare(license_en, ocr_res_dict, field_list): ...@@ -1590,11 +1601,32 @@ def se_mvc34_compare(license_en, ocr_res_dict, field_list):
1590 transfer_num = ocr_res_list[res_idx].get(consts.TRANSFER_NUM, [])[idx] 1601 transfer_num = ocr_res_list[res_idx].get(consts.TRANSFER_NUM, [])[idx]
1591 except Exception as e: 1602 except Exception as e:
1592 transfer_num = empty_str 1603 transfer_num = empty_str
1604 try:
1605 position_info_date = ocr_res_list[res_idx].get(consts.ALL_POSITION_KEY_2, dict()).get(
1606 consts.TRANSFER_DATE, [])[idx]
1607 except Exception as e:
1608 position_info_date = {}
1609 try:
1610 position_info_name = ocr_res_list[res_idx].get(consts.ALL_POSITION_KEY_2, dict()).get(
1611 consts.TRANSFER_NAME, [])[idx]
1612 except Exception as e:
1613 position_info_name = {}
1614 try:
1615 position_info_num = ocr_res_list[res_idx].get(consts.ALL_POSITION_KEY_2, dict()).get(
1616 consts.TRANSFER_NUM, [])[idx]
1617 except Exception as e:
1618 position_info_num = {}
1593 core_info = { 1619 core_info = {
1594 consts.TRANSFER_NAME: transfer_name, 1620 consts.TRANSFER_NAME: transfer_name,
1595 consts.TRANSFER_NUM: transfer_num, 1621 consts.TRANSFER_NUM: transfer_num,
1596 consts.TRANSFER_DATE: transfer_date, 1622 consts.TRANSFER_DATE: transfer_date,
1597 consts.IMG_PATH_KEY: img_path, 1623 consts.IMG_PATH_KEY_2: img_path,
1624 consts.SECTION_IMG_PATH_KEY_2: section_img_path,
1625 consts.ALL_POSITION_KEY: {
1626 consts.TRANSFER_NAME: position_info_name,
1627 consts.TRANSFER_NUM: position_info_num,
1628 consts.TRANSFER_DATE: position_info_date,
1629 },
1598 } 1630 }
1599 page34_date_dict.setdefault(transfer_date, []).append(core_info) 1631 page34_date_dict.setdefault(transfer_date, []).append(core_info)
1600 if first_res is None: 1632 if first_res is None:
...@@ -1613,6 +1645,8 @@ def se_mvc34_compare(license_en, ocr_res_dict, field_list): ...@@ -1613,6 +1645,8 @@ def se_mvc34_compare(license_en, ocr_res_dict, field_list):
1613 if max_date is not None or first_res is not None: 1645 if max_date is not None or first_res is not None:
1614 is_find = True 1646 is_find = True
1615 ocr_res = first_res if max_date is None else page34_date_dict[time.strftime('%Y-%m-%d', max_date)][0] 1647 ocr_res = first_res if max_date is None else page34_date_dict[time.strftime('%Y-%m-%d', max_date)][0]
1648 failed_field = []
1649 base_img_path = ocr_res.get(consts.IMG_PATH_KEY_2, '')
1616 for name, value in field_list: 1650 for name, value in field_list:
1617 ocr_str = ocr_res.get(compare_logic[name][0]) 1651 ocr_str = ocr_res.get(compare_logic[name][0])
1618 if not isinstance(ocr_str, str): 1652 if not isinstance(ocr_str, str):
...@@ -1620,15 +1654,42 @@ def se_mvc34_compare(license_en, ocr_res_dict, field_list): ...@@ -1620,15 +1654,42 @@ def se_mvc34_compare(license_en, ocr_res_dict, field_list):
1620 ocr_str = empty_str 1654 ocr_str = empty_str
1621 else: 1655 else:
1622 result = getattr(cp, compare_logic[name][1])(value, ocr_str, **compare_logic[name][2]) 1656 result = getattr(cp, compare_logic[name][1])(value, ocr_str, **compare_logic[name][2])
1623 img_path = ocr_res.get(consts.IMG_PATH_KEY, '') if result == consts.RESULT_N else empty_str 1657 img_path = base_img_path if result == consts.RESULT_N else empty_str
1624 error_type = empty_error_type if result == consts.RESULT_Y else ErrorType.OCR.value 1658 error_type = empty_error_type if result == consts.RESULT_Y else ErrorType.OCR.value
1625 result_field_list.append((name, value, result, ocr_str, img_path, error_type)) 1659 result_field_list.append((name, value, result, ocr_str, img_path, error_type))
1626 1660
1661 if result == consts.RESULT_N:
1662 failed_field.append(name)
1663
1664 section_img_path = ocr_res.get(consts.SECTION_IMG_PATH_KEY_2, '')
1665 if len(failed_field) > 0 and os.path.exists(section_img_path):
1666 info = ocr_res.get(consts.ALL_POSITION_KEY, {})
1667 try:
1668 last_img = img_process(section_img_path, {}, 0)
1669 except Exception as e:
1670 for field in failed_field:
1671 field_img_path_dict[field] = base_img_path
1672 else:
1673 pre, suf = os.path.splitext(section_img_path)
1674 for field in failed_field:
1675 try:
1676 res_field = compare_logic[field][0]
1677 is_valid, coord_tuple = field_build_coordinates(info.get(res_field, {}))
1678 if is_valid:
1679 save_path = '{0}_{1}{2}'.format(pre, field, suf)
1680 field_img = last_img[coord_tuple[0]:coord_tuple[1], coord_tuple[2]:coord_tuple[3], :]
1681 cv2.imwrite(save_path, field_img)
1682 field_img_path_dict[field] = save_path
1683 else:
1684 field_img_path_dict[field] = base_img_path
1685 except Exception as e:
1686 field_img_path_dict[field] = base_img_path
1687
1627 if not is_find: 1688 if not is_find:
1628 for name, value in field_list: 1689 for name, value in field_list:
1629 result_field_list.append((name, value, consts.RESULT_N, empty_str, empty_str, ErrorType.NF.value)) 1690 result_field_list.append((name, value, consts.RESULT_N, empty_str, empty_str, ErrorType.NF.value))
1630 1691
1631 return result_field_list, dict() 1692 return result_field_list, field_img_path_dict
1632 1693
1633 1694
1634 def se_compare_process(compare_info, ocr_res_dict): 1695 def se_compare_process(compare_info, ocr_res_dict):
......
Styling with Markdown is supported
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!