-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathoutput_data.py
263 lines (237 loc) · 12.8 KB
/
output_data.py
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
import importlib
import supervisely as sly
import yaml
from supervisely.app.widgets import Button, Card, Container, Input, Progress, ProjectThumbnail
g = importlib.import_module("project-dataset.src.globals")
settings = importlib.import_module("project-dataset.src.ui.inference_settings")
inference_preview = importlib.import_module("project-dataset.src.ui.inference_preview")
nn_info = importlib.import_module("project-dataset.src.ui.nn_info")
output_project_name = Input(f"{g.project_info.name}_inference", minlength=1)
apply_button = Button("Apply model to input data", icon="zmdi zmdi-check")
inference_progress = Progress()
output_project_thumbnail = ProjectThumbnail()
output_project_thumbnail.hide()
card = Card(
"6️⃣ Output data",
"New project with predictions will be created. Original project will not be modified.",
content=Container(
[output_project_name, apply_button, inference_progress, output_project_thumbnail]
),
collapsable=True,
lock_message="Connect to the deployed neural network on step 2️⃣.",
)
card.lock()
card.collapse()
def apply_model_ds(src_project, dst_project, inference_settings, res_project_meta):
import time
timer = {}
dst_dataset_infos = []
try:
# 1. Create destination datasets
selected_datasets = g.selected_datasets
dst_dataset_infos = []
dst_image_infos_dict = {} # name -> image_info
src_ds_image_infos_dict = {} # dataset_id -> image_id -> [image_infos]
with inference_progress(
message="Creating datasets...", total=len(selected_datasets)
) as pbar:
src_dataset_infos = g.api.dataset.get_list(src_project)
for src_dataset_info in src_dataset_infos:
t = time.time()
dst_dataset_info = g.api.dataset.copy(
dst_project_id=dst_project.id,
id=src_dataset_info.id,
new_name=src_dataset_info.name,
)
dst_dataset_infos.append(dst_dataset_info)
timer.setdefault(src_dataset_info.id, {})["copy"] = time.time() - t
t = time.time()
for image_info in g.api.image.get_list(dst_dataset_info.id):
dst_image_infos_dict[image_info.name] = image_info
timer.setdefault(src_dataset_info.id, {})["dst_image_infos"] = time.time() - t
t = time.time()
src_ds_image_infos_dict[src_dataset_info.id] = {
image_info.id: image_info
for image_info in g.api.image.get_list(src_dataset_info.id)
}
timer.setdefault(src_dataset_info.id, {})["src_image_infos"] = time.time() - t
pbar.update(1)
# 2. Apply model to the datasets
with inference_progress(message="Processing images...", total=len(g.input_images)) as pbar:
for src_dataset_info in src_dataset_infos:
# iterating over batches of predictions
t = time.time()
for (
_,
merged_ann_infos_batch,
final_project_meta,
) in inference_preview.apply_model_to_datasets(
src_project,
[src_dataset_info.id],
inference_settings,
classes=[
obj_class.name
for obj_class in nn_info.select_classes.get_selected_classes()
],
batch_size=50,
image_infos=list(src_ds_image_infos_dict[src_dataset_info.id].values()),
):
timer.setdefault(src_dataset_info.id, {}).setdefault("items", 0)
timer[src_dataset_info.id]["items"] += len(merged_ann_infos_batch)
timer.setdefault(src_dataset_info.id, {}).setdefault("apply_model", 0)
timer[src_dataset_info.id]["apply_model"] += time.time() - t
t = time.time()
# Update project meta if needed
if res_project_meta != final_project_meta:
res_project_meta = final_project_meta
g.api.project.update_meta(dst_project.id, res_project_meta.to_json())
timer.setdefault(src_dataset_info.id, {}).setdefault("update_meta", 0)
timer[src_dataset_info.id]["update_meta"] += time.time() - t
t = time.time()
dst_anns = []
dst_image_infos = []
for ann_info in merged_ann_infos_batch:
src_image_id = ann_info.image_id
src_image_info = src_ds_image_infos_dict[src_dataset_info.id][src_image_id]
dst_image_infos.append(dst_image_infos_dict[src_image_info.name])
dst_anns.append(
sly.Annotation.from_json(ann_info.annotation, res_project_meta)
)
timer.setdefault(src_dataset_info.id, {}).setdefault("prepare_anns", 0)
timer[src_dataset_info.id]["prepare_anns"] += time.time() - t
t = time.time()
# upload_annotations
try:
g.api.annotation.upload_anns(
[image_info.id for image_info in dst_image_infos], dst_anns
)
pbar.update(len(dst_anns))
except:
for img_info, ann in zip(dst_image_infos, dst_anns):
try:
g.api.annotation.upload_ann(img_info.id, ann)
except Exception as e:
sly.logger.warn(
msg=f"Image: {img_info.name} (Image ID: {img_info.id}) couldn't be uploaded, image will be skipped, error: {e}.",
extra={
"image_name": img_info.name,
"image_id": img_info.id,
"image_meta": img_info.meta,
"image_ann": ann,
},
)
continue
finally:
pbar.update(1)
finally:
timer.setdefault(src_dataset_info.id, {}).setdefault("upload_anns", 0)
timer[src_dataset_info.id]["upload_anns"] += time.time() - t
t = time.time()
except Exception:
g.api.dataset.remove_batch([ds.id for ds in dst_dataset_infos])
raise
finally:
sly.logger.debug("Timer:", extra={"timer": timer})
@apply_button.click
def apply_model():
"""Applies the model to the input data and creates a new project with the predictions.
After the process is finished, the new project will be shown and the app will be stopped."""
try:
inference_settings = yaml.safe_load(settings.additional_settings.get_value())
sly.logger.info(f"Final Inference Settings: {inference_settings}")
except Exception as e:
inference_settings = {}
sly.logger.warning(
f"Model Inference launched without additional settings. \n" f"Reason: {e}",
exc_info=True,
)
res_project_meta = g.project_meta.clone()
res_project = g.api.project.create(
g.workspace_id, output_project_name.get_value(), change_name_if_conflict=True
)
g.api.project.update_meta(res_project.id, res_project_meta.to_json())
# -------------------------------------- Add Workflow Input -------------------------------------- #
g.workflow.add_input(project_id=g.selected_project, session_id=g.model_session_id)
# ----------------------------------------------- - ---------------------------------------------- #
try:
apply_model_ds(g.selected_project, res_project, inference_settings, res_project_meta)
except Exception as e:
sly.logger.warn(
msg=f"Couldn't apply model to the input data, error: {e}.",
exc_info=True,
)
with inference_progress(message="Processing images...", total=len(g.input_images)) as pbar:
for dataset_id in g.selected_datasets:
dataset_info = g.api.dataset.get_info_by_id(dataset_id)
res_dataset = g.api.dataset.create(
res_project.id, dataset_info.name, dataset_info.description
)
image_infos = g.api.image.get_list(dataset_info.id)
for batched_image_infos in sly.batched(image_infos, batch_size=10):
try:
image_ids, res_names, res_metas = [], [], []
for image_info in batched_image_infos:
image_ids.append(image_info.id)
res_names.append(image_info.name)
res_metas.append(image_info.meta)
_, res_anns, final_project_meta = inference_preview.apply_model_to_images(
dataset_info.id, batched_image_infos, inference_settings
)
except Exception as e:
sly.logger.warn(
msg=f"Couldn't process images by batch, images will be processed one by one, error: {e}."
)
image_ids, res_names, res_anns, res_metas = [], [], [], []
for image_info in batched_image_infos:
try:
_, res_ann, final_project_meta = (
inference_preview.apply_model_to_image(
image_info, inference_settings
)
)
image_ids.append(image_info.id)
res_names.append(image_info.name)
res_anns.append(res_ann)
res_metas.append(image_info.meta)
except Exception as e:
sly.logger.warn(
msg=f"Image: {image_info.name} (ID: {image_info.id}) couldn't be processed, image will be skipped, error: {e}.",
extra={
"image_name": image_info.name,
"image_id": image_info.id,
"image_meta": image_info.meta,
},
)
continue
if res_project_meta != final_project_meta:
res_project_meta = final_project_meta
g.api.project.update_meta(res_project.id, res_project_meta.to_json())
res_images_infos = g.api.image.upload_ids(
res_dataset.id, res_names, image_ids, metas=res_metas
)
res_ids = [image_info.id for image_info in res_images_infos]
try:
g.api.annotation.upload_anns(res_ids, res_anns)
except:
for res_img_info, ann in zip(res_images_infos, res_anns):
try:
g.api.annotation.upload_ann(res_img_info.id, ann)
except Exception as e:
sly.logger.warn(
msg=f"Image: {res_img_info.name} (Image ID: {res_img_info.id}) couldn't be uploaded, image will be skipped, error: {e}.",
extra={
"image_name": res_img_info.name,
"image_id": res_img_info.id,
"image_meta": res_img_info.meta,
"image_ann": ann,
},
)
continue
pbar.update(len(batched_image_infos))
output_project_thumbnail.set(g.api.project.get_info_by_id(res_project.id))
output_project_thumbnail.show()
# -------------------------------------- Add Workflow Output ------------------------------------- #
g.workflow.add_output(project_id=res_project.id)
# ----------------------------------------------- - ---------------------------------------------- #
main = importlib.import_module("project-dataset.src.main")
main.app.stop()