Gabriel commited on
Commit
76f8319
1 Parent(s): 1ccc4fd

debug print

Browse files
src/htr_pipeline/inferencer.py CHANGED
@@ -26,25 +26,51 @@ class Inferencer:
26
 
27
  @timer_func
28
  def predict_regions(self, input_image, pred_score_threshold=0.5, containments_threshold=0.5, visualize=True):
 
 
 
 
29
  input_image = self.preprocess_img.binarize_img(input_image)
30
 
31
  image = mmcv.imread(input_image)
 
 
 
 
 
 
 
32
  result = self.seg_model(image, return_datasample=True)
33
  result_pred = result["predictions"][0]
 
 
 
 
 
34
 
35
  filtered_result_pred = self.postprocess_seg_mask.filter_on_pred_threshold(
36
  result_pred, pred_score_threshold=pred_score_threshold
37
  )
38
 
 
 
 
 
39
  if len(filtered_result_pred.pred_instances.masks) == 0:
40
  raise gr.Error("No Regions were predicted by the model")
41
 
42
  else:
 
 
43
  result_align = self.process_seg_mask.align_masks_with_image(filtered_result_pred, image)
44
  result_clean = self.postprocess_seg_mask.remove_overlapping_masks(
45
  predicted_mask=result_align, containments_threshold=containments_threshold
46
  )
47
 
 
 
 
 
48
  if visualize:
49
  result_viz = self.seg_model.visualize(
50
  inputs=[image], preds=[result_clean], return_vis=True, no_save_vis=True
@@ -52,6 +78,8 @@ class Inferencer:
52
  else:
53
  result_viz = None
54
 
 
 
55
  regions_cropped, polygons = self.process_seg_mask.crop_masks(result_clean, image)
56
  order = self.ordering.order_regions_marginalia(result_clean)
57
 
@@ -59,6 +87,10 @@ class Inferencer:
59
  polygons_ordered = [polygons[i] for i in order]
60
  masks_ordered = [result_clean.pred_instances.masks[i] for i in order]
61
 
 
 
 
 
62
  return result_viz, regions_cropped_ordered, polygons_ordered, masks_ordered
63
 
64
  @timer_func
 
26
 
27
  @timer_func
28
  def predict_regions(self, input_image, pred_score_threshold=0.5, containments_threshold=0.5, visualize=True):
29
+ import time
30
+
31
+ t1 = time.time()
32
+
33
  input_image = self.preprocess_img.binarize_img(input_image)
34
 
35
  image = mmcv.imread(input_image)
36
+
37
+ t2 = time.time()
38
+
39
+ print(f"Function executed bin and read in {(t2-t1):.4f}s")
40
+
41
+ t1 = time.time()
42
+
43
  result = self.seg_model(image, return_datasample=True)
44
  result_pred = result["predictions"][0]
45
+ t2 = time.time()
46
+
47
+ print(f"Function executed predict in {(t2-t1):.4f}s")
48
+
49
+ t1 = time.time()
50
 
51
  filtered_result_pred = self.postprocess_seg_mask.filter_on_pred_threshold(
52
  result_pred, pred_score_threshold=pred_score_threshold
53
  )
54
 
55
+ t2 = time.time()
56
+
57
+ print(f"Function executed filter in {(t2-t1):.4f}s")
58
+
59
  if len(filtered_result_pred.pred_instances.masks) == 0:
60
  raise gr.Error("No Regions were predicted by the model")
61
 
62
  else:
63
+ t1 = time.time()
64
+
65
  result_align = self.process_seg_mask.align_masks_with_image(filtered_result_pred, image)
66
  result_clean = self.postprocess_seg_mask.remove_overlapping_masks(
67
  predicted_mask=result_align, containments_threshold=containments_threshold
68
  )
69
 
70
+ t2 = time.time()
71
+
72
+ print(f"Function executed align and remove in {(t2-t1):.4f}s")
73
+
74
  if visualize:
75
  result_viz = self.seg_model.visualize(
76
  inputs=[image], preds=[result_clean], return_vis=True, no_save_vis=True
 
78
  else:
79
  result_viz = None
80
 
81
+ t1 = time.time()
82
+
83
  regions_cropped, polygons = self.process_seg_mask.crop_masks(result_clean, image)
84
  order = self.ordering.order_regions_marginalia(result_clean)
85
 
 
87
  polygons_ordered = [polygons[i] for i in order]
88
  masks_ordered = [result_clean.pred_instances.masks[i] for i in order]
89
 
90
+ t2 = time.time()
91
+
92
+ print(f"Function executed crop and margin in {(t2-t1):.4f}s")
93
+
94
  return result_viz, regions_cropped_ordered, polygons_ordered, masks_ordered
95
 
96
  @timer_func
src/htr_pipeline/utils/pipeline_inferencer.py CHANGED
@@ -6,6 +6,8 @@ from src.htr_pipeline.utils.xml_helper import XMLHelper
6
 
7
  terminate = False
8
 
 
 
9
 
10
  class PipelineInferencer:
11
  def __init__(self, process_seg_mask: SegMaskHelper, xml_helper: XMLHelper):
@@ -29,7 +31,7 @@ class PipelineInferencer:
29
  pred_score_threshold_regions,
30
  pred_score_threshold_lines,
31
  containments_threshold,
32
- htr_threshold=0.7,
33
  ):
34
  global terminate
35
 
@@ -77,7 +79,7 @@ class PipelineInferencer:
77
  return region_data if mean_htr_score > htr_threshold + 0.1 else None
78
 
79
  def _process_lines(
80
- self, text_region, inferencer, pred_score_threshold, containments_threshold, mask, region_id, htr_threshold=0.7
81
  ):
82
  _, lines_cropped_ordered, line_polygons_ordered = inferencer.predict_lines(
83
  text_region, pred_score_threshold, containments_threshold, visualize=False, custom_track=False
 
6
 
7
  terminate = False
8
 
9
+ # TODO check why region is so slow to start.. Is their error with loading the model?
10
+
11
 
12
  class PipelineInferencer:
13
  def __init__(self, process_seg_mask: SegMaskHelper, xml_helper: XMLHelper):
 
31
  pred_score_threshold_regions,
32
  pred_score_threshold_lines,
33
  containments_threshold,
34
+ htr_threshold=0.6,
35
  ):
36
  global terminate
37
 
 
79
  return region_data if mean_htr_score > htr_threshold + 0.1 else None
80
 
81
  def _process_lines(
82
+ self, text_region, inferencer, pred_score_threshold, containments_threshold, mask, region_id, htr_threshold=0.6
83
  ):
84
  _, lines_cropped_ordered, line_polygons_ordered = inferencer.predict_lines(
85
  text_region, pred_score_threshold, containments_threshold, visualize=False, custom_track=False
tabs/htr_tool.py CHANGED
@@ -86,7 +86,7 @@ with gr.Blocks() as htr_tool_tab:
86
  )
87
  with gr.Row():
88
  gr.Slider(
89
- value=0.7,
90
  minimum=0.5,
91
  maximum=1,
92
  label="HTR threshold",
@@ -94,7 +94,7 @@ with gr.Blocks() as htr_tool_tab:
94
  scale=1,
95
  )
96
  gr.Slider(
97
- value=0.8,
98
  minimum=0.6,
99
  maximum=1,
100
  label="Avg threshold",
@@ -105,7 +105,7 @@ with gr.Blocks() as htr_tool_tab:
105
  htr_tool_region_segment_model_dropdown = gr.Dropdown(
106
  choices=["Riksarkivet/RmtDet_region"],
107
  value="Riksarkivet/RmtDet_region",
108
- label="Region segment model",
109
  info="Will add more models later!",
110
  )
111
 
@@ -113,15 +113,15 @@ with gr.Blocks() as htr_tool_tab:
113
  htr_tool_line_segment_model_dropdown = gr.Dropdown(
114
  choices=["Riksarkivet/RmtDet_lines"],
115
  value="Riksarkivet/RmtDet_lines",
116
- label="Line segment model",
117
  info="Will add more models later!",
118
  )
119
 
120
  htr_tool_transcriber_model_dropdown = gr.Dropdown(
121
  choices=["Riksarkivet/SATRN_transcriber", "microsoft/trocr-base-handwritten"],
122
  value="Riksarkivet/SATRN_transcriber",
123
- label="Transcribe model",
124
- info="Will add more models later!",
125
  )
126
 
127
  with gr.Column(scale=2):
 
86
  )
87
  with gr.Row():
88
  gr.Slider(
89
+ value=0.6,
90
  minimum=0.5,
91
  maximum=1,
92
  label="HTR threshold",
 
94
  scale=1,
95
  )
96
  gr.Slider(
97
+ value=0.7,
98
  minimum=0.6,
99
  maximum=1,
100
  label="Avg threshold",
 
105
  htr_tool_region_segment_model_dropdown = gr.Dropdown(
106
  choices=["Riksarkivet/RmtDet_region"],
107
  value="Riksarkivet/RmtDet_region",
108
+ label="Region Segment models",
109
  info="Will add more models later!",
110
  )
111
 
 
113
  htr_tool_line_segment_model_dropdown = gr.Dropdown(
114
  choices=["Riksarkivet/RmtDet_lines"],
115
  value="Riksarkivet/RmtDet_lines",
116
+ label="Line Segment models",
117
  info="Will add more models later!",
118
  )
119
 
120
  htr_tool_transcriber_model_dropdown = gr.Dropdown(
121
  choices=["Riksarkivet/SATRN_transcriber", "microsoft/trocr-base-handwritten"],
122
  value="Riksarkivet/SATRN_transcriber",
123
+ label="Transcriber models",
124
+ info="Models will be continuously updated with future additions for specific cases.",
125
  )
126
 
127
  with gr.Column(scale=2):