File size: 12,685 Bytes
3663121
 
c9a1c2d
 
 
3810c45
c9a1c2d
 
 
 
 
 
089249c
 
c9a1c2d
 
 
 
 
 
 
 
 
 
05085b5
089249c
 
05085b5
089249c
 
c9a1c2d
089249c
05085b5
089249c
 
 
 
3810c45
089249c
 
 
 
 
 
 
05085b5
089249c
451f882
 
 
 
 
 
 
089249c
 
 
 
 
 
6a5ac94
089249c
 
05085b5
b76cf08
 
451f882
 
 
 
 
 
b76cf08
 
 
 
089249c
 
 
b76cf08
 
 
 
 
 
 
 
 
451f882
 
089249c
451f882
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c9a1c2d
089249c
 
c9a1c2d
089249c
 
 
c9a1c2d
b76cf08
69d1982
b76cf08
 
 
 
451f882
b76cf08
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c9a1c2d
089249c
 
c9a1c2d
 
089249c
c9a1c2d
 
 
 
 
 
 
 
089249c
14d4a0b
 
 
 
 
 
 
 
 
089249c
b76cf08
089249c
 
b76cf08
 
 
 
089249c
 
b76cf08
 
089249c
 
 
b76cf08
 
 
 
 
089249c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
60ad418
3663121
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
import os

import gradio as gr

from helper.examples.examples import DemoImages
from helper.utils import TrafficDataHandler
from src.htr_pipeline.gradio_backend import FastTrack, SingletonModelLoader

model_loader = SingletonModelLoader()
fast_track = FastTrack(model_loader)
images_for_demo = DemoImages()

terminate = False


with gr.Blocks() as htr_tool_tab:
    with gr.Row(equal_height=True):
        with gr.Column(scale=2):
            with gr.Row():
                fast_track_input_region_image = gr.Image(
                    label="Image to run HTR on", type="numpy", tool="editor", elem_id="image_upload", height=395
                )

            with gr.Row():
                with gr.Tab("HTRFLOW") as tab_output_and_setting_selector:
                    with gr.Row():
                        stop_htr_button = gr.Button(
                            value="Stop run",
                            variant="stop",
                        )

                        htr_pipeline_button = gr.Button(
                            "Run ",
                            variant="primary",
                            visible=True,
                            elem_id="run_pipeline_button",
                        )
                        htr_pipeline_button_var = gr.State(value="htr_pipeline_button")

                    htr_pipeline_button_api = gr.Button("Run pipeline", variant="primary", visible=False, scale=1)

                    fast_file_downlod = gr.File(
                        label="Download output file", visible=True, scale=1, height=100, elem_id="download_file"
                    )

                with gr.Tab("Visualize") as tab_image_viewer_selector:
                    with gr.Row():
                        gr.Markdown("")
                        #     gr.Button(
                        #     value="Image viewer",
                        #     variant="secondary",
                        #     link="https://huggingface.co/spaces/Riksarkivet/Viewer_demo",
                        #     interactive=True,
                        # )

                        run_image_visualizer_button = gr.Button(
                            value="Visualize results", variant="primary", interactive=True
                        )

                    selection_text_from_image_viewer = gr.Textbox(
                        interactive=False, label="Text Selector", info="Select a line on Image Viewer to return text"
                    )

                with gr.Tab("Compare") as tab_model_compare_selector:
                    with gr.Box():
                        gr.Markdown(
                            """
                            **Work in progress**

                            Compare different runs with uploaded Ground Truth and calculate CER. You will also be able to upload output format files

                            """
                        )

                        calc_cer_button_fast = gr.Button("Calculate CER", variant="primary", visible=True)

        with gr.Column(scale=4):
            with gr.Box():
                with gr.Row(visible=True) as output_and_setting_tab:
                    with gr.Column(scale=2):
                        fast_name_files_placeholder = gr.Markdown(visible=False)
                        gr.Examples(
                            examples=images_for_demo.examples_list,
                            inputs=[fast_name_files_placeholder, fast_track_input_region_image],
                            label="Example images",
                            examples_per_page=5,
                        )

                        gr.Markdown(" ")

                    with gr.Column(scale=3):
                        with gr.Group():
                            gr.Markdown("    ⚙️ Settings ")
                            with gr.Row():
                                radio_file_input = gr.CheckboxGroup(
                                    choices=["Txt", "Page XML"],
                                    value=["Txt", "Page XML"],
                                    label="Output file extension",
                                    info="JSON and ALTO-XML will be added",
                                    scale=1,
                                )
                            with gr.Row():
                                gr.Checkbox(
                                    value=True,
                                    label="Binarize image",
                                    info="Binarize image to reduce background noise",
                                )
                                gr.Checkbox(
                                    value=True,
                                    label="Output prediction threshold",
                                    info="Output XML with prediction score",
                                )

                            with gr.Accordion("Advanced settings", open=False):
                                with gr.Group():
                                    with gr.Row():
                                        htr_tool_region_segment_model_dropdown = gr.Dropdown(
                                            choices=["Riksarkivet/rtmdet_region"],
                                            value="Riksarkivet/rtmdet_region",
                                            label="Region segmentation models",
                                            info="More models will be added",
                                        )

                                        gr.Slider(
                                            minimum=0.4,
                                            maximum=1,
                                            value=0.5,
                                            step=0.05,
                                            label="P-threshold",
                                            info="""Filter confidence score for a prediction score to be considered""",
                                        )

                                    with gr.Row():
                                        htr_tool_line_segment_model_dropdown = gr.Dropdown(
                                            choices=["Riksarkivet/rtmdet_lines"],
                                            value="Riksarkivet/rtmdet_lines",
                                            label="Line segmentation models",
                                            info="More models will be added",
                                        )

                                        gr.Slider(
                                            minimum=0.4,
                                            maximum=1,
                                            value=0.5,
                                            step=0.05,
                                            label="P-threshold",
                                            info="""Filter confidence score for a prediction score to be considered""",
                                        )

                                    with gr.Row():
                                        htr_tool_transcriber_model_dropdown = gr.Dropdown(
                                            choices=["Riksarkivet/satrn_htr", "microsoft/trocr-base-handwritten"],
                                            value="Riksarkivet/satrn_htr",
                                            label="Text recognition models",
                                            info="More models will be added",
                                        )

                                        gr.Slider(
                                            value=0.6,
                                            minimum=0.5,
                                            maximum=1,
                                            label="HTR threshold",
                                            info="Prediction score threshold for transcribed lines",
                                            scale=1,
                                        )
                                    with gr.Row():
                                        gr.Markdown("    More settings will be added")

                with gr.Row(visible=False) as image_viewer_tab:
                    text_polygon_dict = gr.Variable()

                    fast_track_output_image = gr.Image(
                        label="Image Viewer", type="numpy", height=600, interactive=False
                    )

                with gr.Column(visible=False) as model_compare_selector:
                    gr.Markdown("**Work in progress:**")
                    with gr.Row():
                        gr.Radio(
                            choices=["Compare Page XML", "Compare different runs"],
                            value="Compare Page XML",
                            info="Compare different runs from HTRFLOW or with external runs.",
                        )
                    with gr.Row():
                        gr.UploadButton(label="Run A")

                        gr.UploadButton(label="Run B")

                        gr.UploadButton(label="Ground Truth")

                    with gr.Row():
                        gr.HighlightedText(
                            label="Text diff runs",
                            combine_adjacent=True,
                            show_legend=True,
                            color_map={"+": "red", "-": "green"},
                        )

                    with gr.Row():
                        gr.HighlightedText(
                            label="Text diff ground truth",
                            combine_adjacent=True,
                            show_legend=True,
                            color_map={"+": "red", "-": "green"},
                        )

                    with gr.Row():
                        with gr.Column(scale=1):
                            with gr.Row(equal_height=False):
                                cer_output_fast = gr.Textbox(label="CER:")
                        with gr.Column(scale=2):
                            pass

    xml_rendered_placeholder_for_api = gr.Textbox(visible=False)

    htr_event_click_event = htr_pipeline_button.click(
        fast_track.segment_to_xml,
        inputs=[fast_track_input_region_image, radio_file_input],
        outputs=[fast_file_downlod, fast_file_downlod],
    )

    htr_pipeline_button_api.click(
        fast_track.segment_to_xml_api,
        inputs=[fast_track_input_region_image],
        outputs=[xml_rendered_placeholder_for_api],
        api_name="predict",
    )

    def dummy_update_htr_tool_transcriber_model_dropdown(htr_tool_transcriber_model_dropdown):
        return gr.update(value="Riksarkivet/satrn_htr")

    htr_tool_transcriber_model_dropdown.change(
        fn=dummy_update_htr_tool_transcriber_model_dropdown,
        inputs=htr_tool_transcriber_model_dropdown,
        outputs=htr_tool_transcriber_model_dropdown,
    )

    def update_selected_tab_output_and_setting():
        return gr.update(visible=True), gr.update(visible=False), gr.update(visible=False)

    def update_selected_tab_image_viewer():
        return gr.update(visible=False), gr.update(visible=True), gr.update(visible=False)

    def update_selected_tab_model_compare():
        return gr.update(visible=False), gr.update(visible=False), gr.update(visible=True)

    tab_output_and_setting_selector.select(
        fn=update_selected_tab_output_and_setting,
        outputs=[output_and_setting_tab, image_viewer_tab, model_compare_selector],
    )

    tab_image_viewer_selector.select(
        fn=update_selected_tab_image_viewer, outputs=[output_and_setting_tab, image_viewer_tab, model_compare_selector]
    )

    tab_model_compare_selector.select(
        fn=update_selected_tab_model_compare, outputs=[output_and_setting_tab, image_viewer_tab, model_compare_selector]
    )

    def stop_function():
        from src.htr_pipeline.utils import pipeline_inferencer

        pipeline_inferencer.terminate = True
        gr.Info("The HTR execution was halted")

    stop_htr_button.click(fn=stop_function, inputs=None, outputs=None, cancels=[htr_event_click_event])

    run_image_visualizer_button.click(
        fn=fast_track.visualize_image_viewer,
        inputs=fast_track_input_region_image,
        outputs=[fast_track_output_image, text_polygon_dict],
    )

    fast_track_output_image.select(
        fast_track.get_text_from_coords, inputs=text_polygon_dict, outputs=selection_text_from_image_viewer
    )

    SECRET_KEY = os.environ.get("AM_I_IN_A_DOCKER_CONTAINER", False)
    if SECRET_KEY:
        htr_pipeline_button.click(fn=TrafficDataHandler.store_metric_data, inputs=htr_pipeline_button_var)