Spaces:
Running
Running
example image proposed
Browse files- app.py +27 -8
- eval.py +1 -4
- images/example1.jpg +0 -0
- images/example2.jpg +0 -0
- images/example3.jpg +0 -0
- images/example4.jpg +0 -0
- images/none.jpg +0 -0
app.py
CHANGED
@@ -7,6 +7,8 @@ from PIL import Image, ImageEnhance
|
|
7 |
from htlm_webpage import display_bpmn_xml
|
8 |
import gc
|
9 |
import psutil
|
|
|
|
|
10 |
|
11 |
from OCR import text_prediction, filter_text, mapping_text, rescale
|
12 |
from train import prepare_model
|
@@ -22,6 +24,7 @@ from streamlit_image_comparison import image_comparison
|
|
22 |
from xml.dom import minidom
|
23 |
from streamlit_cropper import st_cropper
|
24 |
from streamlit_drawable_canvas import st_canvas
|
|
|
25 |
from utils import find_closest_object
|
26 |
from train import get_faster_rcnn_model, get_arrow_model
|
27 |
import gdown
|
@@ -43,12 +46,13 @@ def read_xml_file(filepath):
|
|
43 |
|
44 |
# Function to modify bounding box positions based on the given sizes
|
45 |
def modif_box_pos(pred, size):
|
46 |
-
|
|
|
47 |
center = [(x1 + x2) / 2, (y1 + y2) / 2]
|
48 |
-
label = class_dict[
|
49 |
if label in size:
|
50 |
-
|
51 |
-
return
|
52 |
|
53 |
# Function to create a BPMN XML file from prediction results
|
54 |
def create_XML(full_pred, text_mapping, scale):
|
@@ -69,7 +73,6 @@ def create_XML(full_pred, text_mapping, scale):
|
|
69 |
'exclusiveGateway': (60, 60),
|
70 |
'event': (43.2, 43.2),
|
71 |
'parallelGateway': (60, 60),
|
72 |
-
'sequenceFlow': (180, 12),
|
73 |
'dataObject': (48, 72),
|
74 |
'dataStore': (72, 72),
|
75 |
'subProcess': (144, 108),
|
@@ -89,7 +92,8 @@ def create_XML(full_pred, text_mapping, scale):
|
|
89 |
})
|
90 |
|
91 |
#modify the boxes positions
|
92 |
-
|
|
|
93 |
|
94 |
# Create BPMN collaboration element
|
95 |
collaboration = ET.SubElement(definitions, 'bpmn:collaboration', id='collaboration_1')
|
@@ -144,6 +148,7 @@ def create_XML(full_pred, text_mapping, scale):
|
|
144 |
pretty_xml_as_string = reparsed.toprettyxml(indent=" ")
|
145 |
|
146 |
full_pred['boxes'] = rescale(1/scale, full_pred['boxes'])
|
|
|
147 |
|
148 |
return pretty_xml_as_string
|
149 |
|
@@ -314,8 +319,22 @@ def main():
|
|
314 |
#Create the layout for the app
|
315 |
col1, col2 = st.columns(2)
|
316 |
with col1:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
317 |
# Create a file uploader for the user to upload an image
|
318 |
-
|
|
|
|
|
|
|
319 |
|
320 |
# Display the uploaded image if the user has uploaded an image
|
321 |
if uploaded_file is not None:
|
@@ -342,7 +361,7 @@ def main():
|
|
342 |
st.session_state.crop_image = cropped_image
|
343 |
with st.spinner('Processing...'):
|
344 |
perform_inference(model_object, model_arrow, st.session_state.crop_image, score_threshold)
|
345 |
-
st.session_state.prediction = modif_box_pos(st.session_state.prediction, object_dict)
|
346 |
st.balloons()
|
347 |
else:
|
348 |
#delete the prediction
|
|
|
7 |
from htlm_webpage import display_bpmn_xml
|
8 |
import gc
|
9 |
import psutil
|
10 |
+
import copy
|
11 |
+
|
12 |
|
13 |
from OCR import text_prediction, filter_text, mapping_text, rescale
|
14 |
from train import prepare_model
|
|
|
24 |
from xml.dom import minidom
|
25 |
from streamlit_cropper import st_cropper
|
26 |
from streamlit_drawable_canvas import st_canvas
|
27 |
+
from streamlit_image_select import image_select
|
28 |
from utils import find_closest_object
|
29 |
from train import get_faster_rcnn_model, get_arrow_model
|
30 |
import gdown
|
|
|
46 |
|
47 |
# Function to modify bounding box positions based on the given sizes
|
48 |
def modif_box_pos(pred, size):
|
49 |
+
modified_pred = copy.deepcopy(pred) # Make a deep copy of the prediction
|
50 |
+
for i, (x1, y1, x2, y2) in enumerate(modified_pred['boxes']):
|
51 |
center = [(x1 + x2) / 2, (y1 + y2) / 2]
|
52 |
+
label = class_dict[modified_pred['labels'][i]]
|
53 |
if label in size:
|
54 |
+
modified_pred['boxes'][i] = [center[0] - size[label][0] / 2, center[1] - size[label][1] / 2, center[0] + size[label][0] / 2, center[1] + size[label][1] / 2]
|
55 |
+
return modified_pred['boxes']
|
56 |
|
57 |
# Function to create a BPMN XML file from prediction results
|
58 |
def create_XML(full_pred, text_mapping, scale):
|
|
|
73 |
'exclusiveGateway': (60, 60),
|
74 |
'event': (43.2, 43.2),
|
75 |
'parallelGateway': (60, 60),
|
|
|
76 |
'dataObject': (48, 72),
|
77 |
'dataStore': (72, 72),
|
78 |
'subProcess': (144, 108),
|
|
|
92 |
})
|
93 |
|
94 |
#modify the boxes positions
|
95 |
+
old_boxes = copy.deepcopy(full_pred)
|
96 |
+
full_pred['boxes'] = modif_box_pos(full_pred, size_elements)
|
97 |
|
98 |
# Create BPMN collaboration element
|
99 |
collaboration = ET.SubElement(definitions, 'bpmn:collaboration', id='collaboration_1')
|
|
|
148 |
pretty_xml_as_string = reparsed.toprettyxml(indent=" ")
|
149 |
|
150 |
full_pred['boxes'] = rescale(1/scale, full_pred['boxes'])
|
151 |
+
full_pred['boxes'] = old_boxes
|
152 |
|
153 |
return pretty_xml_as_string
|
154 |
|
|
|
319 |
#Create the layout for the app
|
320 |
col1, col2 = st.columns(2)
|
321 |
with col1:
|
322 |
+
with st.expander("Use example images"):
|
323 |
+
img_selected = image_select("If you have no image and just want to test the demo, click on one of these images", ["./images/None.jpg", "./images/example1.jpg", "./images/example2.jpg", "./images/example3.jpg"],
|
324 |
+
captions=["None", "Example 1", "Example 2", "Example 3"], index=0, use_container_width=False, return_value="original")
|
325 |
+
|
326 |
+
if img_selected== './images/None.jpg':
|
327 |
+
print('No example image selected')
|
328 |
+
#delete the prediction
|
329 |
+
if 'prediction' in st.session_state:
|
330 |
+
del st.session_state['prediction']
|
331 |
+
img_selected = None
|
332 |
+
|
333 |
# Create a file uploader for the user to upload an image
|
334 |
+
if img_selected is not None:
|
335 |
+
uploaded_file = img_selected
|
336 |
+
else:
|
337 |
+
uploaded_file = st.file_uploader("Choose an image...", type=["jpg", "jpeg", "png"])
|
338 |
|
339 |
# Display the uploaded image if the user has uploaded an image
|
340 |
if uploaded_file is not None:
|
|
|
361 |
st.session_state.crop_image = cropped_image
|
362 |
with st.spinner('Processing...'):
|
363 |
perform_inference(model_object, model_arrow, st.session_state.crop_image, score_threshold)
|
364 |
+
#st.session_state.prediction = modif_box_pos(st.session_state.prediction, object_dict)
|
365 |
st.balloons()
|
366 |
else:
|
367 |
#delete the prediction
|
eval.py
CHANGED
@@ -239,10 +239,7 @@ def create_links(keypoints, boxes, labels, class_dict):
|
|
239 |
if labels[i]==list(class_dict.values()).index('sequenceFlow') or labels[i]==list(class_dict.values()).index('messageFlow'):
|
240 |
closest1, point_start = find_closest_object(keypoints[i][0], boxes, labels)
|
241 |
closest2, point_end = find_closest_object(keypoints[i][1], boxes, labels)
|
242 |
-
|
243 |
-
print('closest1:', closest1, 'closest2:', closest2)
|
244 |
-
print('point_start:', point_start, 'point_end:', point_end)
|
245 |
-
|
246 |
if closest1 is not None and closest2 is not None:
|
247 |
best_points.append([point_start, point_end])
|
248 |
links.append([closest1, closest2])
|
|
|
239 |
if labels[i]==list(class_dict.values()).index('sequenceFlow') or labels[i]==list(class_dict.values()).index('messageFlow'):
|
240 |
closest1, point_start = find_closest_object(keypoints[i][0], boxes, labels)
|
241 |
closest2, point_end = find_closest_object(keypoints[i][1], boxes, labels)
|
242 |
+
|
|
|
|
|
|
|
243 |
if closest1 is not None and closest2 is not None:
|
244 |
best_points.append([point_start, point_end])
|
245 |
links.append([closest1, closest2])
|
images/example1.jpg
ADDED
images/example2.jpg
ADDED
images/example3.jpg
ADDED
images/example4.jpg
ADDED
images/none.jpg
ADDED