Spaces:
Runtime error
Runtime error
SakshiRathi77
commited on
Upload 8 files
Browse files- .gitattributes +2 -0
- app (1).py +296 -0
- app_utils.py +196 -0
- base_line_best_model_exp5.pt +3 -0
- best_sku_model.pt +3 -0
- inference.py +226 -0
- master_tmp.png +3 -0
- requirements (1).txt +40 -0
- to_score_planogram_tmp.png +3 -0
.gitattributes
CHANGED
@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
+
master_tmp.png filter=lfs diff=lfs merge=lfs -text
|
37 |
+
to_score_planogram_tmp.png filter=lfs diff=lfs merge=lfs -text
|
app (1).py
ADDED
@@ -0,0 +1,296 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# https://planogram-compliance.herokuapp.com/
|
2 |
+
# https://dashboard.heroku.com/apps/planogram-compliance/deploy/heroku-git
|
3 |
+
|
4 |
+
# https://medium.com/@mohcufe/how-to-deploy-your-trained-pytorch-model-on-heroku-ff4b73085ddd\
|
5 |
+
# https://stackoverflow.com/questions/51730880/where-do-i-get-a-cpu-only-version-of-pytorch
|
6 |
+
# https://blog.jcharistech.com/2020/02/26/how-to-deploy-a-face-detection-streamlit-app-on-heroku/
|
7 |
+
# https://towardsdatascience.com/a-quick-tutorial-on-how-to-deploy-your-streamlit-app-to-heroku-
|
8 |
+
# https://www.analyticsvidhya.com/blog/2021/06/deploy-your-ml-dl-streamlit-application-on-heroku/
|
9 |
+
# https://gist.github.com/jeremyjordan/6b506257509e8ba673f145baa568a1ea
|
10 |
+
|
11 |
+
import json
|
12 |
+
|
13 |
+
# https://www.r-bloggers.com/2020/12/creating-a-streamlit-web-app-building-with-docker-github-actions-and-hosting-on-heroku/
|
14 |
+
# https://devcenter.heroku.com/articles/container-registry-and-runtime
|
15 |
+
# from yolo_inference_util import run_yolo_v5
|
16 |
+
import os
|
17 |
+
from tempfile import NamedTemporaryFile
|
18 |
+
|
19 |
+
import cv2
|
20 |
+
import numpy as np
|
21 |
+
import pandas as pd
|
22 |
+
import streamlit as st
|
23 |
+
|
24 |
+
# import matplotlib.pyplot as plt
|
25 |
+
from app_utils import annotate_planogram_compliance, bucket_sort, do_sorting, xml_to_csv
|
26 |
+
from inference import run
|
27 |
+
|
28 |
+
# from utils.plots import Annotator, colors
|
29 |
+
# from utils.general import scale_coords
|
30 |
+
|
31 |
+
app_formal_name = "Planogram Compliance"
|
32 |
+
|
33 |
+
FILE_UPLOAD_DIR = "tmp"
|
34 |
+
|
35 |
+
os.makedirs(FILE_UPLOAD_DIR, exist_ok=True)
|
36 |
+
# Start the app in wide-mode
|
37 |
+
st.set_page_config(
|
38 |
+
layout="wide",
|
39 |
+
page_title=app_formal_name,
|
40 |
+
)
|
41 |
+
# https://github.com/streamlit/streamlit/issues/1361
|
42 |
+
uploaded_file = st.file_uploader(
|
43 |
+
"Choose a planogram image to score",
|
44 |
+
type=["jpg", "JPEG", "PNG", "JPG", "jpeg"],
|
45 |
+
)
|
46 |
+
uploaded_master_planogram_file = st.file_uploader(
|
47 |
+
"Upload a master planogram", type=["jpg", "JPEG", "PNG", "JPG", "jpeg"]
|
48 |
+
)
|
49 |
+
annotation_file = st.file_uploader("upload master polanogram", type=["xml"])
|
50 |
+
temp_file = NamedTemporaryFile(delete=False)
|
51 |
+
|
52 |
+
target_names = [
|
53 |
+
"Bottle,100PLUS ACTIVE 1.5L",
|
54 |
+
"Bottle,100PLUS ACTIVE 500ML",
|
55 |
+
"Bottle,100PLUS LEMON LIME 1.5L",
|
56 |
+
"Bottle,100PLUS ORANGE 500ML",
|
57 |
+
"Bottle,100PLUS ORIGINAL 1.5L",
|
58 |
+
"Bottle,100PLUS TANGY ORANGE 1.5L",
|
59 |
+
"Bottle,100PLUS ZERO 1.5L",
|
60 |
+
"Bottle,100PLUS ZERO 500ML",
|
61 |
+
"Packet,F:M MAGNOLIA CHOC 1L",
|
62 |
+
"Bottle,F&N GINGER ADE 1.5L",
|
63 |
+
"Bottle,F&N GRAPE 1.5L",
|
64 |
+
"Bottle,F&N ICE CREAM SODA 1.5L",
|
65 |
+
"Bottle,F&N LYCHEE PEAR 1.5L",
|
66 |
+
"Bottle,F&N ORANGE 1.5L",
|
67 |
+
"Bottle,F&N PINEAPPLE PET 1.5L",
|
68 |
+
"Bottle,F&N SARSI 1.5L",
|
69 |
+
"Bottle,F&N SS ICE LEM TEA RS 500ML",
|
70 |
+
"Bottle,F&N SS ICE LEMON TEA RS 1.5L",
|
71 |
+
"Bottle,F&N SS ICE LEMON TEA 1.5L",
|
72 |
+
"Bottle,F&N SS ICE LEMON TEA 500ML",
|
73 |
+
"Bottle,F&N SS ICE PEACH TEA 1.5L",
|
74 |
+
"Bottle,SS ICE LEMON GT 1.48L",
|
75 |
+
"Bottle,SS WHITE CHRYS TEA 1.48L",
|
76 |
+
"Packet,FARMHOUSE FRESH MILK 1L FNDM",
|
77 |
+
"Packet,FARMHOUSE PLAIN LF 1L",
|
78 |
+
"Packet,PURA FRESH MILK 1L FS",
|
79 |
+
"Packet,NUTRISOY REG NO SUGAR ADDED 1L",
|
80 |
+
"Packet,NUTRISOY PLAIN 475ML",
|
81 |
+
"Packet,NUTRISOY PLAIN 1L",
|
82 |
+
"Packet,NUTRISOY OMEGA RD SUGAR 1L",
|
83 |
+
"Packet,NUTRISOY OMEGA NSA 1L",
|
84 |
+
"Packet,NUTRISOY ALMOND 1L",
|
85 |
+
"Packet,MAGNOLIA FRESH MILK 1L FNDM",
|
86 |
+
"Packet,FM MAG FC PLAIN 200ML",
|
87 |
+
"Packet,MAG OMEGA PLUS PLAIN 200ML",
|
88 |
+
"Packet,MAG KURMA MILK 500ML",
|
89 |
+
"Packet,MAG KURMA MILK 1L",
|
90 |
+
"Packet,MAG CHOCOLATE FC 500ML",
|
91 |
+
"Packet,MAG BROWN SUGAR SS MILK 1L",
|
92 |
+
"Packet,FM MAG LFHC PLN 500ML",
|
93 |
+
"Packet,FM MAG LFHC OAT 500ML",
|
94 |
+
"Packet,FM MAG LFHC OAT 1L",
|
95 |
+
"Packet,FM MAG FC PLAIN 500ML",
|
96 |
+
"Void,PARTIAL VOID",
|
97 |
+
"Void,FULL VOID",
|
98 |
+
"Bottle,F&N SS ICE LEM TEA 500ML",
|
99 |
+
]
|
100 |
+
|
101 |
+
run_app = st.button("Run the compliance check")
|
102 |
+
if run_app and uploaded_file is not None:
|
103 |
+
# Convert the file to an opencv image.
|
104 |
+
file_bytes = np.asarray(bytearray(uploaded_file.read()), dtype=np.uint8)
|
105 |
+
temp_file.write(uploaded_file.getvalue())
|
106 |
+
uploaded_img = cv2.imdecode(file_bytes, 1)
|
107 |
+
cv2.imwrite("tmp/to_score_planogram_tmp.png", uploaded_img)
|
108 |
+
|
109 |
+
# if uploaded_master_planogram_file is None:
|
110 |
+
# master = cv2.imread('./sample_master_planogram.jpeg')
|
111 |
+
|
112 |
+
names_dict = {name: id for id, name in enumerate(target_names)}
|
113 |
+
|
114 |
+
sorted_xml_df = None
|
115 |
+
# https://discuss.streamlit.io/t/unable-to-read-files-using-standard-file-uploader/2258/2
|
116 |
+
if uploaded_master_planogram_file and annotation_file:
|
117 |
+
file_bytes = np.asarray(
|
118 |
+
bytearray(uploaded_master_planogram_file.read()), dtype=np.uint8
|
119 |
+
)
|
120 |
+
master = cv2.imdecode(file_bytes, 1)
|
121 |
+
cv2.imwrite("tmp/master_tmp.png", master)
|
122 |
+
# cv2.imwrite("tmp_uploaded_master_planogram_img.png", master)
|
123 |
+
# xml = annotation_file.read()
|
124 |
+
# tmp_xml ="tmp_xml_annotation.xml"
|
125 |
+
# with open(tmp_xml ,'w',encoding='utf-8') as f:
|
126 |
+
# xml = f.write(xml)
|
127 |
+
xml_df = xml_to_csv(annotation_file)
|
128 |
+
xml_df["cls"] = xml_df["cls"].map(names_dict)
|
129 |
+
sorted_xml_df = do_sorting(xml_df)
|
130 |
+
sorted_xml_df.line_number.value_counts()
|
131 |
+
|
132 |
+
line_data = sorted_xml_df.line_number.value_counts()
|
133 |
+
n_rows = int(len(line_data))
|
134 |
+
n_cols = int(max(line_data))
|
135 |
+
master_table = np.zeros((n_rows, n_cols)) + 101
|
136 |
+
master_annotations = []
|
137 |
+
for i, row in sorted_xml_df.groupby("line_number"):
|
138 |
+
# print(f"Adding products in the row {i} to the detected planogram", row.cls.tolist())
|
139 |
+
products = row.cls.tolist()
|
140 |
+
master_table[int(i - 1), 0 : len(products)] = products
|
141 |
+
annotations = [
|
142 |
+
(int(k), int(v))
|
143 |
+
for k, v in list(
|
144 |
+
zip(row.cls.unique(), row.cls.value_counts().tolist())
|
145 |
+
)
|
146 |
+
]
|
147 |
+
master_annotations.append(annotations)
|
148 |
+
master_table.shape
|
149 |
+
# print("Annoatated planogram")
|
150 |
+
# print(np.matrix(master_table))
|
151 |
+
|
152 |
+
elif uploaded_master_planogram_file:
|
153 |
+
print(
|
154 |
+
"Finding the amster annotations with the YOLOv5 model predictions"
|
155 |
+
)
|
156 |
+
file_bytes = np.asarray(
|
157 |
+
bytearray(uploaded_master_planogram_file.read()), dtype=np.uint8
|
158 |
+
)
|
159 |
+
master = cv2.imdecode(file_bytes, 1)
|
160 |
+
cv2.imwrite("tmp/master_tmp.png", master)
|
161 |
+
master_results = run(
|
162 |
+
weights="base_line_best_model_exp5.pt",
|
163 |
+
source="tmp/master_tmp.png",
|
164 |
+
imgsz=[640, 640],
|
165 |
+
conf_thres=0.6,
|
166 |
+
iou_thres=0.6,
|
167 |
+
)
|
168 |
+
|
169 |
+
bb_df = pd.DataFrame(
|
170 |
+
master_results[0][1].tolist(),
|
171 |
+
columns=["xmin", "ymin", "xmax", "ymax", "conf", "cls"],
|
172 |
+
)
|
173 |
+
sorted_df = do_sorting(bb_df)
|
174 |
+
|
175 |
+
n_rows = int(sorted_df.line_number.max())
|
176 |
+
n_cols = int(
|
177 |
+
sorted_df.groupby("line_number")
|
178 |
+
.size()
|
179 |
+
.reset_index(name="counts")["counts"]
|
180 |
+
.max()
|
181 |
+
)
|
182 |
+
non_null_product = 101
|
183 |
+
print("master size", n_rows, n_cols)
|
184 |
+
master_annotations = []
|
185 |
+
master_table = np.zeros((int(n_rows), int(n_cols))) + non_null_product
|
186 |
+
for i, row in sorted_df.groupby("line_number"):
|
187 |
+
# print(f"Adding products in the row {i} to the detected planogram", row.cls.tolist())
|
188 |
+
products = row.cls.tolist()
|
189 |
+
col_len = min(len(products), n_cols)
|
190 |
+
print("col size: ", col_len)
|
191 |
+
print("row size: ", i - 1)
|
192 |
+
if n_rows <= (i - 1):
|
193 |
+
print("more rows than expected in the predictions")
|
194 |
+
break
|
195 |
+
master_table[int(i - 1), 0:col_len] = products[:col_len]
|
196 |
+
annotations = [
|
197 |
+
(int(k), int(v))
|
198 |
+
for k, v in list(
|
199 |
+
zip(row.cls.unique(), row.cls.value_counts().tolist())
|
200 |
+
)
|
201 |
+
]
|
202 |
+
master_annotations.append(annotations)
|
203 |
+
else:
|
204 |
+
master = cv2.imread("./sample_master_planogram.jpeg")
|
205 |
+
n_rows = 3
|
206 |
+
n_cols = 16
|
207 |
+
master_table = np.zeros((n_rows, n_cols)) + 101
|
208 |
+
master_annotations = [
|
209 |
+
[(32, 12), (8, 4)],
|
210 |
+
[(36, 1), (41, 6), (50, 4), (51, 3), (52, 2)],
|
211 |
+
[(23, 5), (24, 6), (54, 5)],
|
212 |
+
]
|
213 |
+
|
214 |
+
for i, row in enumerate(master_annotations):
|
215 |
+
idx = 0
|
216 |
+
for product, count in row:
|
217 |
+
master_table[i, idx : idx + count] = product
|
218 |
+
idx = idx + count
|
219 |
+
# Now do something with the image! For example, let's display it:
|
220 |
+
# st.image(opencv_image, channels="BGR")
|
221 |
+
|
222 |
+
# uploaded_img = '/content/drive/My Drive/0.CV/0.Planogram_Compliance/planogram_data/images/test/IMG_5718.jpg'
|
223 |
+
result_list = run(
|
224 |
+
weights="base_line_best_model_exp5.pt",
|
225 |
+
source="tmp/to_score_planogram_tmp.png",
|
226 |
+
imgsz=[640, 640],
|
227 |
+
conf_thres=0.6,
|
228 |
+
iou_thres=0.6,
|
229 |
+
)
|
230 |
+
|
231 |
+
bb_df = pd.DataFrame(
|
232 |
+
result_list[0][1].tolist(),
|
233 |
+
columns=["xmin", "ymin", "xmax", "ymax", "conf", "cls"],
|
234 |
+
)
|
235 |
+
sorted_df = do_sorting(bb_df)
|
236 |
+
|
237 |
+
non_null_product = 101
|
238 |
+
print("master size", n_rows, n_cols)
|
239 |
+
detected_table = np.zeros((n_rows, n_cols)) + non_null_product
|
240 |
+
for i, row in sorted_df.groupby("line_number"):
|
241 |
+
# print(f"Adding products in the row {i} to the detected planogram", row.cls.tolist())
|
242 |
+
products = row.cls.tolist()
|
243 |
+
col_len = min(len(products), n_cols)
|
244 |
+
print("col size: ", col_len)
|
245 |
+
print("row size: ", i - 1)
|
246 |
+
if n_rows <= (i - 1):
|
247 |
+
print("more rows than expected in the predictions")
|
248 |
+
break
|
249 |
+
detected_table[int(i - 1), 0:col_len] = products[:col_len]
|
250 |
+
|
251 |
+
# score = (master_table == detected_table).sum() / (master_table != non_null_product).sum()
|
252 |
+
correct_matches = (
|
253 |
+
np.ma.masked_equal(master_table, non_null_product) == detected_table
|
254 |
+
).sum()
|
255 |
+
total_products = (master_table != non_null_product).sum()
|
256 |
+
score = correct_matches / total_products
|
257 |
+
# if sorted_xml_df is not None:
|
258 |
+
# annotate_df = sorted_xml_df[["xmin","ymin", "xmax", "ymax", "line_number","cls"]].astype(int)
|
259 |
+
# else:
|
260 |
+
annotate_df = sorted_df[
|
261 |
+
["xmin", "ymin", "xmax", "ymax", "line_number", "cls"]
|
262 |
+
].astype(int)
|
263 |
+
|
264 |
+
mask = master_table != non_null_product
|
265 |
+
m_detected_table = np.ma.masked_array(master_table, mask=mask)
|
266 |
+
m_annotated_table = np.ma.masked_array(detected_table, mask=mask)
|
267 |
+
|
268 |
+
# wrong_indexes = np.ravel_multi_index(master_table*mask != detected_table*mask, master_table.shape)
|
269 |
+
wrong_indexes = np.where(master_table != detected_table)
|
270 |
+
correct_indexes = np.where(master_table == detected_table)
|
271 |
+
annotated_planogram = annotate_planogram_compliance(
|
272 |
+
uploaded_img, annotate_df, correct_indexes, wrong_indexes, target_names
|
273 |
+
)
|
274 |
+
st.title("Target Products")
|
275 |
+
st.write(json.dumps(target_names))
|
276 |
+
st.title("The master planogram annotation")
|
277 |
+
st.write(
|
278 |
+
"The annotations are based on the index of products from Target products list "
|
279 |
+
)
|
280 |
+
st.write(json.dumps(master_annotations))
|
281 |
+
|
282 |
+
# https://github.com/streamlit/streamlit/issues/888
|
283 |
+
st.image(
|
284 |
+
[master, annotated_planogram, result_list[0][0]],
|
285 |
+
width=512,
|
286 |
+
caption=[
|
287 |
+
"Master planogram",
|
288 |
+
"Planogram Compliance",
|
289 |
+
"Planogram Predictions",
|
290 |
+
],
|
291 |
+
channels="BGR",
|
292 |
+
)
|
293 |
+
# st.image([master, annotated_planogram], width=512, caption=["Master planogram", "Planogram Compliance"], channels="BGR")
|
294 |
+
st.title("Planogram Compiance score")
|
295 |
+
# st.write(f"{correct_matches} / {total_products}")
|
296 |
+
st.write(score)
|
app_utils.py
ADDED
@@ -0,0 +1,196 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import glob
|
2 |
+
import json
|
3 |
+
import os
|
4 |
+
import xml.etree.ElementTree as ET
|
5 |
+
|
6 |
+
import cv2
|
7 |
+
|
8 |
+
# from sklearn.externals import joblib
|
9 |
+
import joblib
|
10 |
+
import numpy as np
|
11 |
+
import pandas as pd
|
12 |
+
|
13 |
+
# from .variables import old_ocr_req_cols
|
14 |
+
# from .skew_correction import PageSkewWraper
|
15 |
+
|
16 |
+
const_HW = 1.294117647
|
17 |
+
const_W = 600
|
18 |
+
# https://www.forbes.com/sites/forbestechcouncil/2020/06/02/leveraging-technologies-to-align-realograms-and-planograms-for-grocery/?sh=506b8b78e86c
|
19 |
+
|
20 |
+
|
21 |
+
# https://stackoverflow.com/questions/39403183/python-opencv-sorting-contours
|
22 |
+
# http://devdoc.net/linux/OpenCV-3.2.0/da/d0c/tutorial_bounding_rects_circles.html
|
23 |
+
# https://stackoverflow.com/questions/10297713/find-contour-of-the-set-of-points-in-opencv
|
24 |
+
# https://stackoverflow.com/questions/16538774/dealing-with-contours-and-bounding-rectangle-in-opencv-2-4-python-2-7
|
25 |
+
# https://stackoverflow.com/questions/50308055/creating-bounding-boxes-for-contours
|
26 |
+
# https://stackoverflow.com/questions/57296398/how-can-i-get-better-results-of-bounding-box-using-find-contours-of-opencv
|
27 |
+
# http://amroamroamro.github.io/mexopencv/opencv/generalContours_demo1.html
|
28 |
+
# https://gist.github.com/bigsnarfdude/d811e31ee17495f82f10db12651ae82d
|
29 |
+
# http://man.hubwiz.com/docset/OpenCV.docset/Contents/Resources/Documents/da/d0c/tutorial_bounding_rects_circles.html
|
30 |
+
# https://www.analyticsvidhya.com/blog/2021/05/document-layout-detection-and-ocr-with-detectron2/
|
31 |
+
# https://colab.research.google.com/drive/1m6gaQF6Q4M0IaSjoo_4jWllKJjK-i6fw?usp=sharing#scrollTo=lEyl3wYKHAe1
|
32 |
+
# https://stackoverflow.com/questions/39403183/python-opencv-sorting-contours
|
33 |
+
# https://docs.opencv.org/2.4/doc/tutorials/imgproc/shapedescriptors/bounding_rects_circles/bounding_rects_circles.html
|
34 |
+
# https://www.pyimagesearch.com/2016/03/21/ordering-coordinates-clockwise-with-python-and-opencv/
|
35 |
+
|
36 |
+
|
37 |
+
def bucket_sort(df, colmn, ymax_col="ymax", ymin_col="ymin"):
|
38 |
+
df["line_number"] = 0
|
39 |
+
colmn.append("line_number")
|
40 |
+
array_value = df[colmn].values
|
41 |
+
start_index = Line_counter = counter = 0
|
42 |
+
ymax, ymin, line_no = (
|
43 |
+
colmn.index(ymax_col),
|
44 |
+
colmn.index(ymin_col),
|
45 |
+
colmn.index("line_number"),
|
46 |
+
)
|
47 |
+
while counter < len(array_value):
|
48 |
+
current_ymax = array_value[start_index][ymax]
|
49 |
+
for next_index in range(start_index, len(array_value)):
|
50 |
+
counter += 1
|
51 |
+
|
52 |
+
next_ymin = array_value[next_index][ymin]
|
53 |
+
next_ymax = array_value[next_index][ymax]
|
54 |
+
if current_ymax > next_ymin:
|
55 |
+
|
56 |
+
array_value[next_index][line_no] = Line_counter + 1
|
57 |
+
# if current_ymax < next_ymax:
|
58 |
+
|
59 |
+
# current_ymax = next_ymax
|
60 |
+
else:
|
61 |
+
counter -= 1
|
62 |
+
break
|
63 |
+
# print(counter, len(array_value), start_index)
|
64 |
+
start_index = counter
|
65 |
+
Line_counter += 1
|
66 |
+
return pd.DataFrame(array_value, columns=colmn)
|
67 |
+
|
68 |
+
|
69 |
+
def do_sorting(df):
|
70 |
+
df.sort_values(["ymin", "xmin"], ascending=True, inplace=True)
|
71 |
+
df["idx"] = df.index
|
72 |
+
if "line_number" in df.columns:
|
73 |
+
print("line number removed")
|
74 |
+
df.drop("line_number", axis=1, inplace=True)
|
75 |
+
req_colns = ["xmin", "ymin", "xmax", "ymax", "idx"]
|
76 |
+
temp_df = df.copy()
|
77 |
+
temp = bucket_sort(temp_df.copy(), req_colns)
|
78 |
+
df = df.merge(temp[["idx", "line_number"]], on="idx")
|
79 |
+
df.sort_values(["line_number", "xmin"], ascending=True, inplace=True)
|
80 |
+
df = df.reset_index(drop=True)
|
81 |
+
df = df.reset_index(drop=True)
|
82 |
+
return df
|
83 |
+
|
84 |
+
|
85 |
+
def xml_to_csv(xml_file):
|
86 |
+
# https://gist.github.com/rotemtam/88d9a4efae243fc77ed4a0f9917c8f6c
|
87 |
+
xml_list = []
|
88 |
+
# for xml_file in glob.glob(path + '/*.xml'):
|
89 |
+
# https://discuss.streamlit.io/t/unable-to-read-files-using-standard-file-uploader/2258/2
|
90 |
+
tree = ET.parse(xml_file)
|
91 |
+
root = tree.getroot()
|
92 |
+
for member in root.findall("object"):
|
93 |
+
bbx = member.find("bndbox")
|
94 |
+
xmin = int(bbx.find("xmin").text)
|
95 |
+
ymin = int(bbx.find("ymin").text)
|
96 |
+
xmax = int(bbx.find("xmax").text)
|
97 |
+
ymax = int(bbx.find("ymax").text)
|
98 |
+
label = member.find("name").text
|
99 |
+
|
100 |
+
value = (
|
101 |
+
root.find("filename").text,
|
102 |
+
int(root.find("size")[0].text),
|
103 |
+
int(root.find("size")[1].text),
|
104 |
+
label,
|
105 |
+
xmin,
|
106 |
+
ymin,
|
107 |
+
xmax,
|
108 |
+
ymax,
|
109 |
+
)
|
110 |
+
xml_list.append(value)
|
111 |
+
column_name = [
|
112 |
+
"filename",
|
113 |
+
"width",
|
114 |
+
"height",
|
115 |
+
"cls",
|
116 |
+
"xmin",
|
117 |
+
"ymin",
|
118 |
+
"xmax",
|
119 |
+
"ymax",
|
120 |
+
]
|
121 |
+
xml_df = pd.DataFrame(xml_list, columns=column_name)
|
122 |
+
return xml_df
|
123 |
+
|
124 |
+
|
125 |
+
# def annotate_planogram_compliance(img0, sorted_xml_df, wrong_indexes, target_names):
|
126 |
+
# # annotator = Annotator(img0, line_width=3, pil=True)
|
127 |
+
# det = sorted_xml_df[['xmin', 'ymin', 'xmax', 'ymax','cls']].values
|
128 |
+
# # det[:, :4] = scale_coords((640, 640), det[:, :4], img0.shape).round()
|
129 |
+
# for i, (*xyxy, cls) in enumerate(det):
|
130 |
+
|
131 |
+
# c = int(cls) # integer class
|
132 |
+
|
133 |
+
# if i in wrong_indexes:
|
134 |
+
# # print(xyxy, "Wrong detection", (255, 0, 0))
|
135 |
+
# label = "Wrong detection"
|
136 |
+
# color = (0,0,255)
|
137 |
+
# else:
|
138 |
+
# # print(xyxy, label, (0, 255, 0))
|
139 |
+
# label = f'{target_names[c]}'
|
140 |
+
# color = (0,255, 0)
|
141 |
+
# org = (int(xyxy[0]), int(xyxy[1]) )
|
142 |
+
# top_left = org
|
143 |
+
# bottom_right = (int(xyxy[2]), int(xyxy[3]))
|
144 |
+
# # print("#"*50)
|
145 |
+
# # print(f"Anooatting cv2 rectangle with shape: { img0.shape}, top left: { top_left}, bottom right: { bottom_right} , color : { color }, thickness: {3}, cv2.LINE_8")
|
146 |
+
# # print("#"*50)
|
147 |
+
# cv2.rectangle(img0, top_left, bottom_right , color, 3, cv2.LINE_8)
|
148 |
+
|
149 |
+
# cv2.putText(img0, label, tuple(org), cv2. FONT_HERSHEY_SIMPLEX , 0.5, color)
|
150 |
+
|
151 |
+
# return img0
|
152 |
+
|
153 |
+
|
154 |
+
def annotate_planogram_compliance(
|
155 |
+
img0, sorted_df, correct_indexes, wrong_indexes, target_names
|
156 |
+
):
|
157 |
+
# annotator = Annotator(img0, line_width=3, pil=True)
|
158 |
+
det = sorted_df[["xmin", "ymin", "xmax", "ymax", "cls"]].values
|
159 |
+
# det[:, :4] = scale_coords((640, 640), det[:, :4], img0.shape).round()
|
160 |
+
for x, y in zip(*correct_indexes):
|
161 |
+
try:
|
162 |
+
row = sorted_df[sorted_df["line_number"] == x + 1].iloc[y]
|
163 |
+
xyxy = row[["xmin", "ymin", "xmax", "ymax"]].values
|
164 |
+
label = f'{target_names[row["cls"]]}'
|
165 |
+
color = (0, 255, 0)
|
166 |
+
# org = (int(xyxy[0]), int(xyxy[1]) )
|
167 |
+
top_left = (int(row["xmin"]), int(row["ymin"]))
|
168 |
+
bottom_right = (int(row["xmax"]), int(row["ymax"]))
|
169 |
+
cv2.rectangle(img0, top_left, bottom_right, color, 3, cv2.LINE_8)
|
170 |
+
|
171 |
+
cv2.putText(
|
172 |
+
img0, label, top_left, cv2.FONT_HERSHEY_SIMPLEX, 0.5, color
|
173 |
+
)
|
174 |
+
except Exception as e:
|
175 |
+
print("Error: " + str(e))
|
176 |
+
continue
|
177 |
+
|
178 |
+
for x, y in zip(*wrong_indexes):
|
179 |
+
try:
|
180 |
+
row = sorted_df[sorted_df["line_number"] == x + 1].iloc[y]
|
181 |
+
xyxy = row[["xmin", "ymin", "xmax", "ymax"]].values
|
182 |
+
label = f'{target_names[row["cls"]]}'
|
183 |
+
color = (0, 0, 255)
|
184 |
+
# org = (int(xyxy[0]), int(xyxy[1]) )
|
185 |
+
top_left = (row["xmin"], row["ymin"])
|
186 |
+
bottom_right = (row["xmax"], row["ymax"])
|
187 |
+
cv2.rectangle(img0, top_left, bottom_right, color, 3, cv2.LINE_8)
|
188 |
+
|
189 |
+
cv2.putText(
|
190 |
+
img0, label, top_left, cv2.FONT_HERSHEY_SIMPLEX, 0.5, color
|
191 |
+
)
|
192 |
+
except Exception as e:
|
193 |
+
print("Error: " + str(e))
|
194 |
+
continue
|
195 |
+
|
196 |
+
return img0
|
base_line_best_model_exp5.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c259d5e97010ee1c9775d6d8c3bc8bb73f52a5ad871ca920902f35563f2acb42
|
3 |
+
size 14621601
|
best_sku_model.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:46627e4923a4cbb695e2f1da5944ec7e2930acb640b822227aab334bddf1548b
|
3 |
+
size 14355573
|
inference.py
ADDED
@@ -0,0 +1,226 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# YOLOv5 🚀 by Ultralytics, GPL-3.0 license
|
2 |
+
"""
|
3 |
+
Run YOLOv5 detection inference on images, videos, directories, globs, YouTube, webcam, streams, etc.
|
4 |
+
|
5 |
+
Usage - sources:
|
6 |
+
$ python detect.py --weights yolov5s.pt --source 0 # webcam
|
7 |
+
img.jpg # image
|
8 |
+
vid.mp4 # video
|
9 |
+
screen # screenshot
|
10 |
+
path/ # directory
|
11 |
+
list.txt # list of images
|
12 |
+
list.streams # list of streams
|
13 |
+
'path/*.jpg' # glob
|
14 |
+
'https://youtu.be/Zgi9g1ksQHc' # YouTube
|
15 |
+
'rtsp://example.com/media.mp4' # RTSP, RTMP, HTTP stream
|
16 |
+
|
17 |
+
Usage - formats:
|
18 |
+
$ python detect.py --weights yolov5s.pt # PyTorch
|
19 |
+
yolov5s.torchscript # TorchScript
|
20 |
+
yolov5s.onnx # ONNX Runtime or OpenCV DNN with --dnn
|
21 |
+
yolov5s_openvino_model # OpenVINO
|
22 |
+
yolov5s.engine # TensorRT
|
23 |
+
yolov5s.mlmodel # CoreML (macOS-only)
|
24 |
+
yolov5s_saved_model # TensorFlow SavedModel
|
25 |
+
yolov5s.pb # TensorFlow GraphDef
|
26 |
+
yolov5s.tflite # TensorFlow Lite
|
27 |
+
yolov5s_edgetpu.tflite # TensorFlow Edge TPU
|
28 |
+
yolov5s_paddle_model # PaddlePaddle
|
29 |
+
"""
|
30 |
+
|
31 |
+
import argparse
|
32 |
+
import os
|
33 |
+
import platform
|
34 |
+
import sys
|
35 |
+
from pathlib import Path
|
36 |
+
|
37 |
+
import torch
|
38 |
+
|
39 |
+
FILE = Path(__file__).resolve()
|
40 |
+
ROOT = FILE.parents[0] # YOLOv5 root directory
|
41 |
+
if str(ROOT) not in sys.path:
|
42 |
+
sys.path.append(str(ROOT)) # add ROOT to PATH
|
43 |
+
ROOT = Path(os.path.relpath(ROOT, Path.cwd())) # relative
|
44 |
+
|
45 |
+
from models.common import DetectMultiBackend
|
46 |
+
from utils.dataloaders import (
|
47 |
+
IMG_FORMATS,
|
48 |
+
VID_FORMATS,
|
49 |
+
LoadImages,
|
50 |
+
LoadScreenshots,
|
51 |
+
LoadStreams,
|
52 |
+
)
|
53 |
+
from utils.general import (
|
54 |
+
LOGGER,
|
55 |
+
Profile,
|
56 |
+
check_file,
|
57 |
+
check_img_size,
|
58 |
+
check_imshow,
|
59 |
+
check_requirements,
|
60 |
+
colorstr,
|
61 |
+
cv2,
|
62 |
+
increment_path,
|
63 |
+
non_max_suppression,
|
64 |
+
print_args,
|
65 |
+
scale_boxes,
|
66 |
+
strip_optimizer,
|
67 |
+
xyxy2xywh,
|
68 |
+
)
|
69 |
+
from utils.plots import Annotator, colors, save_one_box
|
70 |
+
from utils.torch_utils import select_device, smart_inference_mode
|
71 |
+
|
72 |
+
|
73 |
+
@smart_inference_mode()
|
74 |
+
def run(
|
75 |
+
weights=ROOT / "yolov5s.pt", # model path or triton URL
|
76 |
+
source=ROOT / "data/images", # file/dir/URL/glob/screen/0(webcam)
|
77 |
+
data=ROOT / "data/coco128.yaml", # dataset.yaml path
|
78 |
+
imgsz=(640, 640), # inference size (height, width)
|
79 |
+
conf_thres=0.25, # confidence threshold
|
80 |
+
iou_thres=0.45, # NMS IOU threshold
|
81 |
+
max_det=1000, # maximum detections per image
|
82 |
+
device="", # cuda device, i.e. 0 or 0,1,2,3 or cpu
|
83 |
+
view_img=False, # show results
|
84 |
+
save_txt=False, # save results to *.txt
|
85 |
+
save_conf=False, # save confidences in --save-txt labels
|
86 |
+
save_crop=False, # save cropped prediction boxes
|
87 |
+
nosave=False, # do not save images/videos
|
88 |
+
classes=None, # filter by class: --class 0, or --class 0 2 3
|
89 |
+
agnostic_nms=False, # class-agnostic NMS
|
90 |
+
augment=False, # augmented inference
|
91 |
+
visualize=False, # visualize features
|
92 |
+
update=False, # update all models
|
93 |
+
project=ROOT / "runs/detect", # save results to project/name
|
94 |
+
name="exp", # save results to project/name
|
95 |
+
exist_ok=False, # existing project/name ok, do not increment
|
96 |
+
line_thickness=3, # bounding box thickness (pixels)
|
97 |
+
hide_labels=False, # hide labels
|
98 |
+
hide_conf=False, # hide confidences
|
99 |
+
half=False, # use FP16 half-precision inference
|
100 |
+
dnn=False, # use OpenCV DNN for ONNX inference
|
101 |
+
vid_stride=1, # video frame-rate stride
|
102 |
+
):
|
103 |
+
source = str(source)
|
104 |
+
save_img = not nosave and not source.endswith(
|
105 |
+
".txt"
|
106 |
+
) # save inference images
|
107 |
+
is_file = Path(source).suffix[1:] in (IMG_FORMATS + VID_FORMATS)
|
108 |
+
is_url = source.lower().startswith(
|
109 |
+
("rtsp://", "rtmp://", "http://", "https://")
|
110 |
+
)
|
111 |
+
webcam = (
|
112 |
+
source.isnumeric()
|
113 |
+
or source.endswith(".streams")
|
114 |
+
or (is_url and not is_file)
|
115 |
+
)
|
116 |
+
screenshot = source.lower().startswith("screen")
|
117 |
+
if is_url and is_file:
|
118 |
+
source = check_file(source) # download
|
119 |
+
|
120 |
+
# Directories
|
121 |
+
save_dir = increment_path(
|
122 |
+
Path(project) / name, exist_ok=exist_ok
|
123 |
+
) # increment run
|
124 |
+
(save_dir / "labels" if save_txt else save_dir).mkdir(
|
125 |
+
parents=True, exist_ok=True
|
126 |
+
) # make dir
|
127 |
+
|
128 |
+
# Load model
|
129 |
+
device = select_device(device)
|
130 |
+
model = DetectMultiBackend(
|
131 |
+
weights, device=device, dnn=dnn, data=data, fp16=half
|
132 |
+
)
|
133 |
+
stride, names, pt = model.stride, model.names, model.pt
|
134 |
+
imgsz = check_img_size(imgsz, s=stride) # check image size
|
135 |
+
|
136 |
+
# Dataloader
|
137 |
+
bs = 1 # batch_size
|
138 |
+
if webcam:
|
139 |
+
view_img = check_imshow(warn=True)
|
140 |
+
dataset = LoadStreams(
|
141 |
+
source,
|
142 |
+
img_size=imgsz,
|
143 |
+
stride=stride,
|
144 |
+
auto=pt,
|
145 |
+
vid_stride=vid_stride,
|
146 |
+
)
|
147 |
+
bs = len(dataset)
|
148 |
+
elif screenshot:
|
149 |
+
dataset = LoadScreenshots(
|
150 |
+
source, img_size=imgsz, stride=stride, auto=pt
|
151 |
+
)
|
152 |
+
else:
|
153 |
+
dataset = LoadImages(
|
154 |
+
source,
|
155 |
+
img_size=imgsz,
|
156 |
+
stride=stride,
|
157 |
+
auto=pt,
|
158 |
+
vid_stride=vid_stride,
|
159 |
+
)
|
160 |
+
vid_path, vid_writer = [None] * bs, [None] * bs
|
161 |
+
|
162 |
+
# Run inference
|
163 |
+
model.warmup(imgsz=(1 if pt or model.triton else bs, 3, *imgsz)) # warmup
|
164 |
+
seen, windows, dt = 0, [], (Profile(), Profile(), Profile())
|
165 |
+
for path, im, im0s, vid_cap, s in dataset:
|
166 |
+
with dt[0]:
|
167 |
+
im = torch.from_numpy(im).to(model.device)
|
168 |
+
im = im.half() if model.fp16 else im.float() # uint8 to fp16/32
|
169 |
+
im /= 255 # 0 - 255 to 0.0 - 1.0
|
170 |
+
if len(im.shape) == 3:
|
171 |
+
im = im[None] # expand for batch dim
|
172 |
+
|
173 |
+
# Inference
|
174 |
+
with dt[1]:
|
175 |
+
visualize = (
|
176 |
+
increment_path(save_dir / Path(path).stem, mkdir=True)
|
177 |
+
if visualize
|
178 |
+
else False
|
179 |
+
)
|
180 |
+
pred = model(im, augment=augment, visualize=visualize)
|
181 |
+
|
182 |
+
# NMS
|
183 |
+
with dt[2]:
|
184 |
+
pred = non_max_suppression(
|
185 |
+
pred,
|
186 |
+
conf_thres,
|
187 |
+
iou_thres,
|
188 |
+
classes,
|
189 |
+
agnostic_nms,
|
190 |
+
max_det=max_det,
|
191 |
+
)
|
192 |
+
|
193 |
+
# Second-stage classifier (optional)
|
194 |
+
# pred = utils.general.apply_classifier(pred, classifier_model, im, im0s)
|
195 |
+
|
196 |
+
# Process predictions
|
197 |
+
for i, det in enumerate(pred): # per image
|
198 |
+
seen += 1
|
199 |
+
if webcam: # batch_size >= 1
|
200 |
+
p, im0, frame = path[i], im0s[i].copy(), dataset.count
|
201 |
+
s += f"{i}: "
|
202 |
+
else:
|
203 |
+
p, im0, frame = path, im0s.copy(), getattr(dataset, "frame", 0)
|
204 |
+
|
205 |
+
p = Path(p) # to Path
|
206 |
+
save_path = str(save_dir / p.name) # im.jpg
|
207 |
+
txt_path = str(save_dir / "labels" / p.stem) + (
|
208 |
+
"" if dataset.mode == "image" else f"_{frame}"
|
209 |
+
) # im.txt
|
210 |
+
s += "%gx%g " % im.shape[2:] # print string
|
211 |
+
gn = torch.tensor(im0.shape)[
|
212 |
+
[1, 0, 1, 0]
|
213 |
+
] # normalization gain whwh
|
214 |
+
imc = im0.copy() if save_crop else im0 # for save_crop
|
215 |
+
annotator = Annotator(
|
216 |
+
im0, line_width=line_thickness, example=str(names)
|
217 |
+
)
|
218 |
+
results = []
|
219 |
+
if len(det):
|
220 |
+
# Rescale boxes from img_size to im0 size
|
221 |
+
det[:, :4] = scale_boxes(
|
222 |
+
im.shape[2:], det[:, :4], im0.shape
|
223 |
+
).round()
|
224 |
+
results.append((path, det))
|
225 |
+
|
226 |
+
return results
|
master_tmp.png
ADDED
Git LFS Details
|
requirements (1).txt
ADDED
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
# pip install -r requirements.txt
|
3 |
+
# Base ----------------------------------------
|
4 |
+
seaborn
|
5 |
+
ipython
|
6 |
+
psutil
|
7 |
+
matplotlib>=3.2.2
|
8 |
+
numpy>=1.18.5
|
9 |
+
opencv-python==4.7.0.68
|
10 |
+
Pillow>=7.1.2
|
11 |
+
PyYAML>=5.3.1
|
12 |
+
requests>=2.23.0
|
13 |
+
scipy>=1.4.1
|
14 |
+
torch>=1.7.0
|
15 |
+
torchvision>=0.8.1
|
16 |
+
tqdm>=4.41.0
|
17 |
+
joblib
|
18 |
+
# Logging -------------------------------------
|
19 |
+
#tensorboard>=2.4.1
|
20 |
+
#wandb
|
21 |
+
|
22 |
+
# Plotting ------------------------------------
|
23 |
+
pandas>=1.1.4
|
24 |
+
#seaborn>=0.11.0
|
25 |
+
|
26 |
+
# Export --------------------------------------
|
27 |
+
# coremltools>=4.1 # CoreML export
|
28 |
+
# onnx>=1.9.0 # ONNX export
|
29 |
+
# onnx-simplifier>=0.3.6 # ONNX simplifier
|
30 |
+
# scikit-learn==0.19.2 # CoreML quantization
|
31 |
+
# tensorflow>=2.4.1 # TFLite export
|
32 |
+
# tensorflowjs>=3.9.0 # TF.js export
|
33 |
+
# openvino-dev # OpenVINO export
|
34 |
+
|
35 |
+
# Extras --------------------------------------
|
36 |
+
# albumentations>=1.0.3
|
37 |
+
# Cython # for pycocotools https://github.com/cocodataset/cocoapi/issues/172
|
38 |
+
# pycocotools>=2.0 # COCO mAP
|
39 |
+
# roboflow
|
40 |
+
#thop # FLOPs computation
|
to_score_planogram_tmp.png
ADDED
Git LFS Details
|