mineru 2.6.0__py3-none-any.whl → 2.6.2__py3-none-any.whl

mineru/backend/vlm/utils.py

@@ -44,7 +44,7 @@ def enable_custom_logits_processors() -> bool:
         return True
 
 
-def set_defult_gpu_memory_utilization() -> float:
+def set_default_gpu_memory_utilization() -> float:
     from vllm import __version__ as vllm_version
     if version.parse(vllm_version) >= version.parse("0.11.0"):
         return 0.7
@@ -52,7 +52,7 @@ def set_defult_gpu_memory_utilization() -> float:
         return 0.5
 
 
-def set_defult_batch_size() -> int:
+def set_default_batch_size() -> int:
     try:
         device = get_device()
         vram = get_vram(device)

mineru/backend/vlm/vlm_analyze.py

@@ -4,7 +4,7 @@ import time
 
 from loguru import logger
 
-from .utils import enable_custom_logits_processors, set_defult_gpu_memory_utilization, set_defult_batch_size
+from .utils import enable_custom_logits_processors, set_default_gpu_memory_utilization, set_default_batch_size
 from .model_output_to_middle_json import result_to_middle_json
 from ...data.data_reader_writer import DataWriter
 from mineru.utils.pdf_image_tools import load_images_from_pdf
@@ -74,9 +74,11 @@ class ModelSingleton:
                         use_fast=True,
                     )
                     if batch_size == 0:
-                        batch_size = set_defult_batch_size()
+                        batch_size = set_default_batch_size()
                 else:
-                    os.environ["OMP_NUM_THREADS"] = "1"
+                    if os.getenv('OMP_NUM_THREADS') is None:
+                        os.environ["OMP_NUM_THREADS"] = "1"
+
                     if backend == "vllm-engine":
                         try:
                             import vllm
@@ -84,7 +86,7 @@ class ModelSingleton:
                         except ImportError:
                             raise ImportError("Please install vllm to use the vllm-engine backend.")
                         if "gpu_memory_utilization" not in kwargs:
-                            kwargs["gpu_memory_utilization"] = set_defult_gpu_memory_utilization()
+                            kwargs["gpu_memory_utilization"] = set_default_gpu_memory_utilization()
                         if "model" not in kwargs:
                             kwargs["model"] = model_path
                         if enable_custom_logits_processors() and ("logits_processors" not in kwargs):
@@ -99,7 +101,7 @@ class ModelSingleton:
                         except ImportError:
                             raise ImportError("Please install vllm to use the vllm-async-engine backend.")
                         if "gpu_memory_utilization" not in kwargs:
-                            kwargs["gpu_memory_utilization"] = set_defult_gpu_memory_utilization()
+                            kwargs["gpu_memory_utilization"] = set_default_gpu_memory_utilization()
                         if "model" not in kwargs:
                             kwargs["model"] = model_path
                         if enable_custom_logits_processors() and ("logits_processors" not in kwargs):

mineru/model/utils/pytorchocr/utils/resources/arch_config.yaml

@@ -28,82 +28,6 @@ Multilingual_PP-OCRv3_det_infer:
     name: DBHead
     k: 50
 
-en_PP-OCRv3_det_infer:
-  model_type: det
-  algorithm: DB
-  Transform:
-  Backbone:
-    name: MobileNetV3
-    scale: 0.5
-    model_name: large
-    disable_se: True
-  Neck:
-    name: RSEFPN
-    out_channels: 96
-    shortcut: True
-  Head:
-    name: DBHead
-    k: 50
-
-ch_PP-OCRv3_det_infer:
-  model_type: det
-  algorithm: DB
-  Transform:
-  Backbone:
-    name: MobileNetV3
-    scale: 0.5
-    model_name: large
-    disable_se: True
-  Neck:
-    name: RSEFPN
-    out_channels: 96
-    shortcut: True
-  Head:
-    name: DBHead
-    k: 50
-
-en_PP-OCRv4_rec_infer:
-  model_type: rec
-  algorithm: SVTR_LCNet
-  Transform:
-  Backbone:
-    name: PPLCNetV3
-    scale: 0.95
-  Head:
-    name: MultiHead
-    out_channels_list:
-      CTCLabelDecode: 97 #'blank' + ...(62) + ' '
-    head_list:
-      - CTCHead:
-          Neck:
-            name: svtr
-            dims: 120
-            depth: 2
-            hidden_dims: 120
-            kernel_size: [ 1, 3 ]
-            use_guide: True
-          Head:
-            fc_decay: 0.00001
-      - NRTRHead:
-          nrtr_dim: 384
-          max_text_length: 25
-
-ch_PP-OCRv4_det_infer:
-  model_type: det
-  algorithm: DB
-  Transform: null
-  Backbone:
-    name: PPLCNetV3
-    scale: 0.75
-    det: True
-  Neck:
-    name: RSEFPN
-    out_channels: 96
-    shortcut: True
-  Head:
-    name: DBHead
-    k: 50
-
 ch_PP-OCRv5_det_infer:
   model_type: det
   algorithm: DB
@@ -152,57 +76,6 @@ ch_PP-OCRv4_det_server_infer:
     k: 50
     mode: "large"
 
-ch_PP-OCRv4_rec_infer:
-  model_type: rec
-  algorithm: SVTR_LCNet
-  Transform:
-  Backbone:
-    name: PPLCNetV3
-    scale: 0.95
-  Head:
-    name: MultiHead
-    out_channels_list:
-      CTCLabelDecode: 6625 #'blank' + ...(6623) + ' '
-    head_list:
-      - CTCHead:
-          Neck:
-            name: svtr
-            dims: 120
-            depth: 2
-            hidden_dims: 120
-            kernel_size: [ 1, 3 ]
-            use_guide: True
-          Head:
-            fc_decay: 0.00001
-      - NRTRHead:
-          nrtr_dim: 384
-          max_text_length: 25
-
-ch_PP-OCRv4_rec_server_infer:
-  model_type: rec
-  algorithm: SVTR_HGNet
-  Transform:
-  Backbone:
-    name: PPHGNet_small
-  Head:
-    name: MultiHead
-    out_channels_list:
-      CTCLabelDecode: 6625 #'blank' + ...(6623) + ' '
-    head_list:
-      - CTCHead:
-          Neck:
-            name: svtr
-            dims: 120
-            depth: 2
-            hidden_dims: 120
-            kernel_size: [ 1, 3 ]
-            use_guide: True
-          Head:
-            fc_decay: 0.00001
-      - NRTRHead:
-          nrtr_dim: 384
-          max_text_length: 25
-
 ch_PP-OCRv4_rec_server_doc_infer:
   model_type: rec
   algorithm: SVTR_HGNet
@@ -280,174 +153,6 @@ ch_PP-OCRv5_rec_infer:
           nrtr_dim: 384
           max_text_length: 25
 
-chinese_cht_PP-OCRv3_rec_infer:
-  model_type: rec
-  algorithm: SVTR
-  Transform:
-  Backbone:
-    name: MobileNetV1Enhance
-    scale: 0.5
-    last_conv_stride: [1, 2]
-    last_pool_type: avg
-  Neck:
-    name: SequenceEncoder
-    encoder_type: svtr
-    dims: 64
-    depth: 2
-    hidden_dims: 120
-    use_guide: True
-  Head:
-    name: CTCHead
-#    out_channels: 8423
-    fc_decay: 0.00001
-
-latin_PP-OCRv3_rec_infer:
-  model_type: rec
-  algorithm: SVTR
-  Transform:
-  Backbone:
-    name: MobileNetV1Enhance
-    scale: 0.5
-    last_conv_stride: [ 1, 2 ]
-    last_pool_type: avg
-  Neck:
-    name: SequenceEncoder
-    encoder_type: svtr
-    dims: 64
-    depth: 2
-    hidden_dims: 120
-    use_guide: True
-  Head:
-    name: CTCHead
-#    out_channels: 187
-    fc_decay: 0.00001
-
-cyrillic_PP-OCRv3_rec_infer:
-  model_type: rec
-  algorithm: SVTR
-  Transform:
-  Backbone:
-    name: MobileNetV1Enhance
-    scale: 0.5
-    last_conv_stride: [ 1, 2 ]
-    last_pool_type: avg
-  Neck:
-    name: SequenceEncoder
-    encoder_type: svtr
-    dims: 64
-    depth: 2
-    hidden_dims: 120
-    use_guide: True
-  Head:
-    name: CTCHead
-#    out_channels: 165
-    fc_decay: 0.00001
-
-arabic_PP-OCRv3_rec_infer:
-  model_type: rec
-  algorithm: SVTR
-  Transform:
-  Backbone:
-    name: MobileNetV1Enhance
-    scale: 0.5
-    last_conv_stride: [ 1, 2 ]
-    last_pool_type: avg
-  Neck:
-    name: SequenceEncoder
-    encoder_type: svtr
-    dims: 64
-    depth: 2
-    hidden_dims: 120
-    use_guide: True
-  Head:
-    name: CTCHead
-#    out_channels: 164
-    fc_decay: 0.00001
-
-korean_PP-OCRv3_rec_infer:
-  model_type: rec
-  algorithm: SVTR
-  Transform:
-  Backbone:
-    name: MobileNetV1Enhance
-    scale: 0.5
-    last_conv_stride: [ 1, 2 ]
-    last_pool_type: avg
-  Neck:
-    name: SequenceEncoder
-    encoder_type: svtr
-    dims: 64
-    depth: 2
-    hidden_dims: 120
-    use_guide: True
-  Head:
-    name: CTCHead
-#    out_channels: 3690
-    fc_decay: 0.00001
-
-japan_PP-OCRv3_rec_infer:
-  model_type: rec
-  algorithm: SVTR
-  Transform:
-  Backbone:
-    name: MobileNetV1Enhance
-    scale: 0.5
-    last_conv_stride: [ 1, 2 ]
-    last_pool_type: avg
-  Neck:
-    name: SequenceEncoder
-    encoder_type: svtr
-    dims: 64
-    depth: 2
-    hidden_dims: 120
-    use_guide: True
-  Head:
-    name: CTCHead
-#    out_channels: 4401
-    fc_decay: 0.00001
-
-ta_PP-OCRv3_rec_infer:
-  model_type: rec
-  algorithm: SVTR
-  Transform:
-  Backbone:
-    name: MobileNetV1Enhance
-    scale: 0.5
-    last_conv_stride: [ 1, 2 ]
-    last_pool_type: avg
-  Neck:
-    name: SequenceEncoder
-    encoder_type: svtr
-    dims: 64
-    depth: 2
-    hidden_dims: 120
-    use_guide: True
-  Head:
-    name: CTCHead
-#    out_channels: 130
-    fc_decay: 0.00001
-
-te_PP-OCRv3_rec_infer:
-  model_type: rec
-  algorithm: SVTR
-  Transform:
-  Backbone:
-    name: MobileNetV1Enhance
-    scale: 0.5
-    last_conv_stride: [ 1, 2 ]
-    last_pool_type: avg
-  Neck:
-    name: SequenceEncoder
-    encoder_type: svtr
-    dims: 64
-    depth: 2
-    hidden_dims: 120
-    use_guide: True
-  Head:
-    name: CTCHead
-#    out_channels: 153
-    fc_decay: 0.00001
-
 ka_PP-OCRv3_rec_infer:
   model_type: rec
   algorithm: SVTR
@@ -469,27 +174,6 @@ ka_PP-OCRv3_rec_infer:
 #    out_channels: 155
     fc_decay: 0.00001
 
-devanagari_PP-OCRv3_rec_infer:
-  model_type: rec
-  algorithm: SVTR
-  Transform:
-  Backbone:
-    name: MobileNetV1Enhance
-    scale: 0.5
-    last_conv_stride: [ 1, 2 ]
-    last_pool_type: avg
-  Neck:
-    name: SequenceEncoder
-    encoder_type: svtr
-    dims: 64
-    depth: 2
-    hidden_dims: 120
-    use_guide: True
-  Head:
-    name: CTCHead
-#    out_channels: 169
-    fc_decay: 0.00001
-
 korean_PP-OCRv5_rec_infer:
   model_type: rec
   algorithm: SVTR_HGNet
@@ -526,7 +210,7 @@ latin_PP-OCRv5_rec_infer:
   Head:
     name: MultiHead
     out_channels_list:
-      CTCLabelDecode: 504
+      CTCLabelDecode: 838
     head_list:
       - CTCHead:
           Neck: