openocr-python 0.0.2__py3-none-any.whl

openocr/configs/rec/visionlan/resnet45_trans_visionlan_LF_1.yml

@@ -0,0 +1,102 @@
+Global:
+  device: gpu
+  epoch_num: 10
+  log_smooth_window: 20
+  print_batch_step: 10
+  output_dir: ./output/rec/u14m_filter/resnet45_trans_visionlan_LF1/
+  eval_epoch_step: [0, 1]
+  eval_batch_step: [0, 500]
+  cal_metric_during_train: True
+  pretrained_model:
+  checkpoints:
+  use_tensorboard: false
+  infer_img:
+  # for data or label process
+  character_dict_path: &character_dict_path ./tools/utils/EN_symbol_dict.txt # 96en
+  # ./tools/utils/ppocr_keys_v1.txt  # ch
+  max_text_length: &max_text_length 25
+  use_space_char: &use_space_char False
+  save_res_path: ./output/rec/u14m_filter/predicts_resnet45_trans_visionlan_LF1.txt
+  grad_clip_val: 20
+  use_amp: True
+
+Optimizer:
+  name: Adam
+  lr: 0.0002 # for 4gpus bs128/gpu
+  weight_decay: 0.0
+
+LRScheduler:
+  name: MultiStepLR
+  milestones: [6]
+
+Architecture:
+  model_type: rec
+  algorithm: VisionLAN
+  Transform:
+  Encoder:
+    name: ResNet45
+    in_channels: 3
+    strides: [2, 2, 2, 1, 1]
+  Decoder:
+    name: VisionLANDecoder
+    training_step: &training_step 'LF_1'
+    n_position: 256
+
+Loss:
+  name: VisionLANLoss
+  training_step: *training_step
+
+PostProcess:
+  name: VisionLANLabelDecode
+  character_dict_path: *character_dict_path
+  use_space_char: *use_space_char
+
+Metric:
+  name: RecMetric
+  main_indicator: acc
+  is_filter: True
+
+Train:
+  dataset:
+    name: LMDBDataSet
+    data_dir: ../Union14M-L-LMDB-Filtered
+    transforms:
+      - DecodeImagePIL: # load image
+          img_mode: RGB
+      - PARSeqAugPIL:
+      - VisionLANLabelEncode:
+          character_dict_path: *character_dict_path
+          use_space_char: *use_space_char
+          max_text_length: *max_text_length
+      - RecTVResize:
+          image_shape: [64, 256]
+          padding: False
+      - KeepKeys:
+          keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order
+  loader:
+    shuffle: True
+    batch_size_per_card: 128
+    drop_last: True
+    num_workers: 4
+
+Eval:
+  dataset:
+    name: LMDBDataSet
+    data_dir: ../evaluation
+    transforms:
+      - DecodeImagePIL: # load image
+          img_mode: RGB
+      - VisionLANLabelEncode:
+          character_dict_path: *character_dict_path
+          use_space_char: *use_space_char
+          max_text_length: *max_text_length
+      - RecTVResize:
+          image_shape: [64, 256]
+          padding: False
+      - KeepKeys:
+          keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order
+  loader:
+    shuffle: False
+    drop_last: False
+    batch_size_per_card: 128
+    num_workers: 2

openocr/configs/rec/visionlan/resnet45_trans_visionlan_LF_2.yml

@@ -0,0 +1,103 @@
+Global:
+  device: gpu
+  epoch_num: 10
+  log_smooth_window: 20
+  print_batch_step: 10
+  output_dir: ./output/rec/u14m_filter/resnet45_trans_visionlan_LF2/
+  eval_epoch_step: [0, 1]
+  eval_batch_step: [0, 500]
+  cal_metric_during_train: True
+  pretrained_model:
+  # ./output/rec/u14m_filter/resnet45_trans_visionlan_LF1/best.pth
+  checkpoints:
+  use_tensorboard: false
+  infer_img:
+  # for data or label process
+  character_dict_path: &character_dict_path ./tools/utils/EN_symbol_dict.txt # 96en
+  # ./tools/utils/ppocr_keys_v1.txt  # ch
+  max_text_length: &max_text_length 25
+  use_space_char: &use_space_char False
+  save_res_path: ./output/rec/u14m_filter/predicts_resnet45_trans_visionlan_LF2.txt
+  grad_clip_val: 20
+  use_amp: True
+
+Optimizer:
+  name: Adam
+  lr: 0.0002 # for 4gpus bs128/gpu
+  weight_decay: 0.0
+
+LRScheduler:
+  name: MultiStepLR
+  milestones: [6]
+
+Architecture:
+  model_type: rec
+  algorithm: VisionLAN
+  Transform:
+  Encoder:
+    name: ResNet45
+    in_channels: 3
+    strides: [2, 2, 2, 1, 1]
+  Decoder:
+    name: VisionLANDecoder
+    training_step: &training_step 'LF_2'
+    n_position: 256
+
+Loss:
+  name: VisionLANLoss
+  training_step: *training_step
+
+PostProcess:
+  name: VisionLANLabelDecode
+  character_dict_path: *character_dict_path
+  use_space_char: *use_space_char
+
+Metric:
+  name: RecMetric
+  main_indicator: acc
+  is_filter: True
+
+Train:
+  dataset:
+    name: LMDBDataSet
+    data_dir: ../Union14M-L-LMDB-Filtered
+    transforms:
+      - DecodeImagePIL: # load image
+          img_mode: RGB
+      - PARSeqAugPIL:
+      - VisionLANLabelEncode:
+          character_dict_path: *character_dict_path
+          use_space_char: *use_space_char
+          max_text_length: *max_text_length
+      - RecTVResize:
+          image_shape: [64, 256]
+          padding: False
+      - KeepKeys:
+          keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order
+  loader:
+    shuffle: True
+    batch_size_per_card: 128
+    drop_last: True
+    num_workers: 4
+
+Eval:
+  dataset:
+    name: LMDBDataSet
+    data_dir: ../evaluation
+    transforms:
+      - DecodeImagePIL: # load image
+          img_mode: RGB
+      - VisionLANLabelEncode:
+          character_dict_path: *character_dict_path
+          use_space_char: *use_space_char
+          max_text_length: *max_text_length
+      - RecTVResize:
+          image_shape: [64, 256]
+          padding: False
+      - KeepKeys:
+          keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order
+  loader:
+    shuffle: False
+    drop_last: False
+    batch_size_per_card: 128
+    num_workers: 2

openocr/configs/rec/visionlan/svtrv2_visionlan_LA.yml

@@ -0,0 +1,112 @@
+Global:
+  device: gpu
+  epoch_num: 20
+  log_smooth_window: 20
+  print_batch_step: 10
+  output_dir: ./output/rec/u14m_filter/svtrv2_visionlan_LA/
+  eval_epoch_step: [0, 1]
+  eval_batch_step: [0, 500]
+  cal_metric_during_train: True
+  pretrained_model:
+  # ./output/rec/u14m_filter/svtrv2_visionlan_LF2/best.pth
+  checkpoints:
+  use_tensorboard: false
+  infer_img:
+  # for data or label process
+  character_dict_path: &character_dict_path ./tools/utils/EN_symbol_dict.txt # 96en
+  # ./tools/utils/ppocr_keys_v1.txt  # ch
+  max_text_length: &max_text_length 25
+  use_space_char: &use_space_char False
+  save_res_path: ./output/rec/u14m_filter/predicts_svtrv2_visionlan_LA.txt
+  grad_clip_val: 20
+  use_amp: True
+
+Optimizer:
+  name: AdamW
+  lr: 0.00065 # for 4gpus bs256/gpu
+  weight_decay: 0.05
+  filter_bias_and_bn: True
+
+LRScheduler:
+  name: OneCycleLR
+  warmup_epoch: 1.5 # pct_start 0.075*20 = 1.5ep
+  cycle_momentum: False
+
+Architecture:
+  model_type: rec
+  algorithm: VisionLAN
+  Transform:
+  Encoder:
+    name: SVTRv2LNConvTwo33
+    use_pos_embed: False
+    dims: [128, 256, 384]
+    depths: [6, 6, 6]
+    num_heads: [4, 8, 12]
+    mixer: [['Conv','Conv','Conv','Conv','Conv','Conv'],['Conv','Conv','FGlobal','Global','Global','Global'],['Global','Global','Global','Global','Global','Global']]
+    local_k: [[5, 5], [5, 5], [-1, -1]]
+    sub_k: [[1, 1], [2, 1], [-1, -1]]
+    last_stage: false
+    feat2d: True
+  Decoder:
+    name: VisionLANDecoder
+    training_step: &training_step 'LA'
+    n_position: 128
+
+Loss:
+  name: VisionLANLoss
+  training_step: *training_step
+
+PostProcess:
+  name: VisionLANLabelDecode
+  character_dict_path: *character_dict_path
+  use_space_char: *use_space_char
+
+Metric:
+  name: RecMetric
+  main_indicator: acc
+  is_filter: True
+
+Train:
+  dataset:
+    name: LMDBDataSet
+    data_dir: ../Union14M-L-LMDB-Filtered
+    transforms:
+      - DecodeImagePIL: # load image
+          img_mode: RGB
+      - PARSeqAugPIL:
+      - VisionLANLabelEncode:
+          character_dict_path: *character_dict_path
+          use_space_char: *use_space_char
+          max_text_length: *max_text_length
+      - RecTVResize:
+          image_shape: [32, 128]
+          padding: False
+      - KeepKeys:
+          keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order
+  loader:
+    shuffle: True
+    batch_size_per_card: 256
+    drop_last: True
+    num_workers: 4
+
+Eval:
+  dataset:
+    name: LMDBDataSet
+    data_dir: ../evaluation
+    transforms:
+      - DecodeImagePIL: # load image
+          img_mode: RGB
+      - VisionLANLabelEncode:
+          character_dict_path: *character_dict_path
+          use_space_char: *use_space_char
+          max_text_length: *max_text_length
+      - RecTVResize:
+          image_shape: [32, 128]
+          padding: False
+      - KeepKeys:
+          keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order
+  loader:
+    shuffle: False
+    drop_last: False
+    batch_size_per_card: 256
+    num_workers: 2

openocr/configs/rec/visionlan/svtrv2_visionlan_LF_1.yml

@@ -0,0 +1,111 @@
+Global:
+  device: gpu
+  epoch_num: 10
+  log_smooth_window: 20
+  print_batch_step: 10
+  output_dir: ./output/rec/u14m_filter/svtrv2_visionlan_LF1/
+  eval_epoch_step: [0, 1]
+  eval_batch_step: [0, 500]
+  cal_metric_during_train: True
+  pretrained_model:
+  checkpoints:
+  use_tensorboard: false
+  infer_img:
+  # for data or label process
+  character_dict_path: &character_dict_path ./tools/utils/EN_symbol_dict.txt # 96en
+  # ./tools/utils/ppocr_keys_v1.txt  # ch
+  max_text_length: &max_text_length 25
+  use_space_char: &use_space_char False
+  save_res_path: ./output/rec/u14m_filter/predicts_svtrv2_visionlan_LF1.txt
+  grad_clip_val: 20
+  use_amp: True
+
+Optimizer:
+  name: AdamW
+  lr: 0.00065 # for 4gpus bs256/gpu
+  weight_decay: 0.05
+  filter_bias_and_bn: True
+
+LRScheduler:
+  name: OneCycleLR
+  warmup_epoch: 1.5 # pct_start 0.075*20 = 1.5ep
+  cycle_momentum: False
+
+Architecture:
+  model_type: rec
+  algorithm: VisionLAN
+  Transform:
+  Encoder:
+    name: SVTRv2LNConvTwo33
+    use_pos_embed: False
+    dims: [128, 256, 384]
+    depths: [6, 6, 6]
+    num_heads: [4, 8, 12]
+    mixer: [['Conv','Conv','Conv','Conv','Conv','Conv'],['Conv','Conv','FGlobal','Global','Global','Global'],['Global','Global','Global','Global','Global','Global']]
+    local_k: [[5, 5], [5, 5], [-1, -1]]
+    sub_k: [[1, 1], [2, 1], [-1, -1]]
+    last_stage: false
+    feat2d: True
+  Decoder:
+    name: VisionLANDecoder
+    training_step: &training_step 'LF_1'
+    n_position: 128
+
+Loss:
+  name: VisionLANLoss
+  training_step: *training_step
+
+PostProcess:
+  name: VisionLANLabelDecode
+  character_dict_path: *character_dict_path
+  use_space_char: *use_space_char
+
+Metric:
+  name: RecMetric
+  main_indicator: acc
+  is_filter: True
+
+Train:
+  dataset:
+    name: LMDBDataSet
+    data_dir: ../Union14M-L-LMDB-Filtered
+    transforms:
+      - DecodeImagePIL: # load image
+          img_mode: RGB
+      - PARSeqAugPIL:
+      - VisionLANLabelEncode:
+          character_dict_path: *character_dict_path
+          use_space_char: *use_space_char
+          max_text_length: *max_text_length
+      - RecTVResize:
+          image_shape: [32, 128]
+          padding: False
+      - KeepKeys:
+          keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order
+  loader:
+    shuffle: True
+    batch_size_per_card: 256
+    drop_last: True
+    num_workers: 4
+
+Eval:
+  dataset:
+    name: LMDBDataSet
+    data_dir: ../evaluation
+    transforms:
+      - DecodeImagePIL: # load image
+          img_mode: RGB
+      - VisionLANLabelEncode:
+          character_dict_path: *character_dict_path
+          use_space_char: *use_space_char
+          max_text_length: *max_text_length
+      - RecTVResize:
+          image_shape: [32, 128]
+          padding: False
+      - KeepKeys:
+          keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order
+  loader:
+    shuffle: False
+    drop_last: False
+    batch_size_per_card: 256
+    num_workers: 2

openocr/configs/rec/visionlan/svtrv2_visionlan_LF_2.yml

@@ -0,0 +1,112 @@
+Global:
+  device: gpu
+  epoch_num: 10
+  log_smooth_window: 20
+  print_batch_step: 10
+  output_dir: ./output/rec/u14m_filter/svtrv2_visionlan_LF2/
+  eval_epoch_step: [0, 1]
+  eval_batch_step: [0, 500]
+  cal_metric_during_train: True
+  pretrained_model:
+  # ./output/rec/u14m_filter/svtrv2_visionlan_LF1/best.pth
+  checkpoints:
+  use_tensorboard: false
+  infer_img:
+  # for data or label process
+  character_dict_path: &character_dict_path ./tools/utils/EN_symbol_dict.txt # 96en
+  # ./tools/utils/ppocr_keys_v1.txt  # ch
+  max_text_length: &max_text_length 25
+  use_space_char: &use_space_char False
+  save_res_path: ./output/rec/u14m_filter/predicts_svtrv2_visionlan_LF2.txt
+  grad_clip_val: 20
+  use_amp: True
+
+Optimizer:
+  name: AdamW
+  lr: 0.00065 # for 4gpus bs256/gpu
+  weight_decay: 0.05
+  filter_bias_and_bn: True
+
+LRScheduler:
+  name: OneCycleLR
+  warmup_epoch: 1 # pct_start 0.075*20 = 1.5ep
+  cycle_momentum: False
+
+Architecture:
+  model_type: rec
+  algorithm: VisionLAN
+  Transform:
+  Encoder:
+    name: SVTRv2LNConvTwo33
+    use_pos_embed: False
+    dims: [128, 256, 384]
+    depths: [6, 6, 6]
+    num_heads: [4, 8, 12]
+    mixer: [['Conv','Conv','Conv','Conv','Conv','Conv'],['Conv','Conv','FGlobal','Global','Global','Global'],['Global','Global','Global','Global','Global','Global']]
+    local_k: [[5, 5], [5, 5], [-1, -1]]
+    sub_k: [[1, 1], [2, 1], [-1, -1]]
+    last_stage: false
+    feat2d: True
+  Decoder:
+    name: VisionLANDecoder
+    training_step: &training_step 'LF_2'
+    n_position: 128
+
+Loss:
+  name: VisionLANLoss
+  training_step: *training_step
+
+PostProcess:
+  name: VisionLANLabelDecode
+  character_dict_path: *character_dict_path
+  use_space_char: *use_space_char
+
+Metric:
+  name: RecMetric
+  main_indicator: acc
+  is_filter: True
+
+Train:
+  dataset:
+    name: LMDBDataSet
+    data_dir: ../Union14M-L-LMDB-Filtered
+    transforms:
+      - DecodeImagePIL: # load image
+          img_mode: RGB
+      - PARSeqAugPIL:
+      - VisionLANLabelEncode:
+          character_dict_path: *character_dict_path
+          use_space_char: *use_space_char
+          max_text_length: *max_text_length
+      - RecTVResize:
+          image_shape: [32, 128]
+          padding: False
+      - KeepKeys:
+          keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order
+  loader:
+    shuffle: True
+    batch_size_per_card: 256
+    drop_last: True
+    num_workers: 4
+
+Eval:
+  dataset:
+    name: LMDBDataSet
+    data_dir: ../evaluation
+    transforms:
+      - DecodeImagePIL: # load image
+          img_mode: RGB
+      - VisionLANLabelEncode:
+          character_dict_path: *character_dict_path
+          use_space_char: *use_space_char
+          max_text_length: *max_text_length
+      - RecTVResize:
+          image_shape: [32, 128]
+          padding: False
+      - KeepKeys:
+          keep_keys: ['image', 'label', 'label_res', 'label_sub', 'label_id', 'length'] # dataloader will return list in this order
+  loader:
+    shuffle: False
+    drop_last: False
+    batch_size_per_card: 256
+    num_workers: 2

openocr/demo_gradio.py

@@ -0,0 +1,128 @@
+import os
+import gradio as gr  # gradio==4.20.0
+
+os.environ['FLAGS_allocator_strategy'] = 'auto_growth'
+import cv2
+import numpy as np
+import json
+import time
+from PIL import Image
+from tools.infer_e2e import OpenOCR, check_and_download_font, draw_ocr_box_txt
+
+drop_score = 0.01
+text_sys = OpenOCR(drop_score=drop_score)
+# warm up 5 times
+if True:
+    img = np.random.uniform(0, 255, [640, 640, 3]).astype(np.uint8)
+    for i in range(5):
+        res = text_sys(img_numpy=img)
+font_path = './simfang.ttf'
+check_and_download_font(font_path)
+
+
+def main(input_image):
+    img = input_image[:, :, ::-1]
+    starttime = time.time()
+    results, time_dict, mask = text_sys(img_numpy=img, return_mask=True)
+    elapse = time.time() - starttime
+    save_pred = json.dumps(results[0], ensure_ascii=False)
+    image = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
+    boxes = [res['points'] for res in results[0]]
+    txts = [res['transcription'] for res in results[0]]
+    scores = [res['score'] for res in results[0]]
+    draw_img = draw_ocr_box_txt(
+        image,
+        boxes,
+        txts,
+        scores,
+        drop_score=drop_score,
+        font_path=font_path,
+    )
+    mask = mask[0, 0, :, :] > 0.3
+    return save_pred, elapse, draw_img, mask.astype('uint8') * 255
+
+
+def get_all_file_names_including_subdirs(dir_path):
+    all_file_names = []
+
+    for root, dirs, files in os.walk(dir_path):
+        for file_name in files:
+            all_file_names.append(os.path.join(root, file_name))
+
+    file_names_only = [os.path.basename(file) for file in all_file_names]
+    return file_names_only
+
+
+def list_image_paths(directory):
+    image_extensions = ('.png', '.jpg', '.jpeg', '.gif', '.bmp', '.tiff')
+
+    image_paths = []
+
+    for root, dirs, files in os.walk(directory):
+        for file in files:
+            if file.lower().endswith(image_extensions):
+                relative_path = os.path.relpath(os.path.join(root, file),
+                                                directory)
+                full_path = os.path.join(directory, relative_path)
+                image_paths.append(full_path)
+
+    return image_paths
+
+
+def find_file_in_current_dir_and_subdirs(file_name):
+    for root, dirs, files in os.walk('.'):
+        if file_name in files:
+            relative_path = os.path.join(root, file_name)
+            return relative_path
+
+
+def predict1(input_image, Model_type, OCR_type):
+    if OCR_type == 'E2E':
+        return 11111, 'E2E', input_image
+    elif OCR_type == 'STR':
+        return 11111, 'STR', input_image
+    else:
+        return 11111, 'STD', input_image
+
+
+e2e_img_example = list_image_paths('./OCR_e2e_img')
+
+if __name__ == '__main__':
+    css = '.image-container img { width: 100%; max-height: 320px;}'
+
+    with gr.Blocks(css=css) as demo:
+        gr.HTML("""
+                <h1 style='text-align: center;'>OpenOCR</h1>""")
+        with gr.Row():
+            with gr.Column(scale=1):
+                input_image = gr.Image(label='Input image',
+                                       elem_classes=['image-container'])
+
+                examples = gr.Examples(examples=e2e_img_example,
+                                       inputs=input_image,
+                                       label='Examples')
+                downstream = gr.Button('Run')
+
+            with gr.Column(scale=1):
+                img_mask = gr.Image(label='mask',
+                                    interactive=False,
+                                    elem_classes=['image-container'])
+                img_output = gr.Image(label=' ',
+                                      interactive=False,
+                                      elem_classes=['image-container'])
+
+                output = gr.Textbox(label='Result')
+                confidence = gr.Textbox(label='Latency')
+
+            downstream.click(fn=main,
+                             inputs=[
+                                 input_image,
+                             ],
+                             outputs=[
+                                 output,
+                                 confidence,
+                                 img_output,
+                                 img_mask,
+                             ])
+
+    demo.launch(share=True)

openocr/opendet/modeling/__init__.py

@@ -0,0 +1,11 @@
+import copy
+
+from .base_detector import BaseDetector
+
+__all__ = ['build_model']
+
+
+def build_model(config):
+    config = copy.deepcopy(config)
+    det_model = BaseDetector(config)
+    return det_model

openocr/opendet/modeling/backbones/__init__.py

@@ -0,0 +1,14 @@
+__all__ = ['build_backbone']
+
+
+def build_backbone(config):
+    # det backbone
+    from .repvit import RepSVTR_det
+
+    support_dict = ['RepSVTR_det']
+
+    module_name = config.pop('name')
+    assert module_name in support_dict, Exception(
+        'head only support {}'.format(support_dict))
+    module_class = eval(module_name)(**config)
+    return module_class