-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathjsonReader_flowers.py
182 lines (163 loc) · 12.4 KB
/
jsonReader_flowers.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
# -*- coding: utf-8 -*-
"""
Created on Wed Sep 30 00:56:55 2020
@author: CrisO
"""
#import packages
import json
import csv
import os
from shapely.geometry import Polygon, LineString
from PIL import Image
#set up ditctionaries
#Relation between the precalculated label mappings for each imagefile in the dataset (csv) and the actual metadata of the dataset with the coordinates of the labels (json)
metadataRelations = {
"Ambrosiana_annotations.csv":"Ambrosiana_exported_finished_json.json" ,
"Artstor_annotations.csv":"Artstor_exported_finished_json.json",
"Ashmolean_annotations.csv":"Ashmolean_exported_finished_json.json",
#"BeniCulturali_annotations.csv":"BeniCulturaliModified_exported_finished_json.json", (There where some issues with the BeniCulturali Dataset. Because of deadlines the dataset was left out)
"Boijmans_annotations.csv":"Bojimans_exported_finished_json.json",
"British-Libraries_annotations.csv":"British_Libraries_exported_finished_json.json",
"Cini_annotations.csv":"Cini_exported_finished_json.json",
"Joconde_II_annotations.csv":"Joconde_exported_finished_json.json",
"NGA_annotations.csv":"NGA_exported_finished_json.json",
"RKD_annotations.csv":"RKD_exported_finished_json.json",
"SLUB_annotations.csv":"SLUB_exported_finished_json.json",
"STÄDEL_annotations.csv":"Staedel_exported_finished_json.json",
"wga_annotations.csv":"wga_exported_finished_json.json"
}
#Relation between the datasets and the folder names in which the associated images are stored
pictureRelations = {
"Ambrosiana_annotations.csv": "Ambrosiana",
"Artstor_annotations.csv":"Artstor",
"Ashmolean_annotations.csv":"Ashmolean",
#"BeniCulturali_annotations.csv":"BeniCulturali",
"Boijmans_annotations.csv":"Boijmans",
"British-Libraries_annotations.csv":"British Libraries",
"Cini_annotations.csv":"Cini",
"Joconde_II_annotations.csv":"Joconde_II",
"NGA_annotations.csv":"_NGA",
"RKD_annotations.csv":"RKD Auswahl",
"SLUB_annotations.csv":"SLUB",
"STÄDEL_annotations.csv":"Städel II",
"wga_annotations.csv":"wga"
}
# custom dict structure for temporary storage of files, images and associated metadata per dataset
datasetDict_flower = {
"ambrosiana_flowers" : {"files": [], "csv_Metadata": "Ambrosiana_annotations.csv"},
"artstor_flowers" : {"files": [], "csv_Metadata": "Artstor_annotations.csv"},
"ashmolean_flowers" : {"files": [], "csv_Metadata": "Ashmolean_annotations.csv"},
#"beniCulturali_flowers" : {"files": [], "csv_Metadata": "BeniCulturali_annotations.csv"},
"boijmans_flowers" : {"files": [], "csv_Metadata": "Boijmans_annotations.csv"},
"britishLibraries_flowers" : {"files": [], "csv_Metadata": "British-Libraries_annotations.csv"},
"cini_flowers" : {"files": [], "csv_Metadata": "Cini_annotations.csv"},
"joconde_flowers" : {"files": [], "csv_Metadata": "Joconde_II_annotations.csv"},
"nga_flowers" : {"files": [], "csv_Metadata": "NGA_annotations.csv"},
"rkd_flowers" : {"files": [], "csv_Metadata": "RKD_annotations.csv"},
"slub_flowers" : {"files": [], "csv_Metadata": "SLUB_annotations.csv"},
"staedel_flowers" : {"files": [], "csv_Metadata": "STÄDEL_annotations.csv"},
"wga_flowers" : {"files": [], "csv_Metadata": "wga_annotations.csv"},
}
os.chdir("pathxy") #change directory to folder where results from csvReader.py are stored
#get all imagefiles, sort them according to associated dataset and save them in dict
with open("flowers.csv", newline="") as csvfile:
flowers = csv.DictReader(csvfile, delimiter=",") #open generated csv file flowers.csv from csvReader.py
for row in flowers: #loop through all entries in csv, sort them according to associated dataset and save them in dict
if(row["dataset"] == "Ambrosiana_annotations.csv"):
datasetDict_flower["ambrosiana_flowers"]["files"].append({"filename": row["image_file"], "regions" :[]})
elif(row["dataset"] == "Artstor_annotations.csv"):
datasetDict_flower["artstor_flowers"]["files"].append({"filename":row["image_file"], "regions" :[]})
elif(row["dataset"] == "Ashmolean_annotations.csv"):
datasetDict_flower["ashmolean_flowers"]["files"].append({"filename":row["image_file"], "regions" :[]})
#elif(row["dataset"] == "BeniCulturali_annotations.csv"):
# datasetDict_flower["beniCulturali_flowers"]["files"].append({"filename":row["image_file"], "regions" :[]})
elif(row["dataset"] == "Boijmans_annotations.csv"):
datasetDict_flower["boijmans_flowers"]["files"].append({"filename":row["image_file"], "regions" :[]})
elif(row["dataset"] == "British-Libraries_annotations.csv"):
datasetDict_flower["britishLibraries_flowers"]["files"].append({"filename":row["image_file"], "regions" :[]})
elif(row["dataset"] == "Cini_annotations.csv"):
datasetDict_flower["cini_flowers"]["files"].append({"filename":row["image_file"], "regions" :[]})
elif(row["dataset"] == "Joconde_II_annotations.csv"):
datasetDict_flower["joconde_flowers"]["files"].append({"filename":row["image_file"], "regions" :[]})
elif(row["dataset"] == "NGA_annotations.csv"):
datasetDict_flower["nga_flowers"]["files"].append({"filename":row["image_file"], "regions" :[]})
elif(row["dataset"] == "RKD_annotations.csv"):
datasetDict_flower["rkd_flowers"]["files"].append({"filename":row["image_file"], "regions" :[]})
elif(row["dataset"] == "SLUB_annotations.csv"):
datasetDict_flower["slub_flowers"]["files"].append({"filename":row["image_file"], "regions" :[]})
elif(row["dataset"] == "STÄDEL_annotations.csv"):
datasetDict_flower["staedel_flowers"]["files"].append({"filename":row["image_file"], "regions" :[]})
elif(row["dataset"] == "wga_annotations.csv"):
datasetDict_flower["wga_flowers"]["files"].append({"filename":row["image_file"], "regions" :[]})
else:
print("No matching Dataset found: ", row["dataset"])
os.chdir("pathxy") #change directory to folder where annotation metadata is saved
#get all regions from the according json files tagged with "flower"
#convert polygon coordinates to rect coordinates
#save them in dict
for values in datasetDict_flower.values(): #loop through all datasets and their associated values
with open(metadataRelations[values["csv_Metadata"]], "r") as read_file: # for each dataset: open associated json metadata file
data = json.load(read_file) #convert json in python dict
for file in values["files"]: #loop through all associated image files for the current dataset
for entry in data.values(): #loop through all entries in the metadata (there is one entry for every annotated image file)
if(entry["filename"] == file["filename"]): #If filenames of current image file and metadata entry are the same
for region in entry["regions"]: #loop through all annotated regions in this specific entry for the image file
if(region["region_attributes"]["object/attribute/figure"] == "flower"): #if region has the wished label (flower)
if(region["shape_attributes"]["name"] == "polygon"): #if the region has the shape of a polygon (Polygons cannot be handled in later steps so they have to be converted into rects)
polytemp = [];
for i in range(len(region["shape_attributes"]["all_points_x"])): #get all points of the polygon
coordtemp = [region["shape_attributes"]["all_points_x"][i], region["shape_attributes"]["all_points_y"][i]]
polytemp.append(coordtemp) #save them in a list
polygon = Polygon(polytemp) #create a python polygon object out of them
poly_bbox = polygon.bounds #get the bounding box (a list of the coordinates of the surrounding rectangle) of this newly created polygon
x = poly_bbox[0] #save associated x value of bounding box
y = poly_bbox[1] #save associated y value of bounding box
width = poly_bbox[2] - poly_bbox[0] #calculate width out of bounding box coordinates
height = poly_bbox[3] - poly_bbox[1] #calculate height out of bounding box coordinates
rectRegion = { #create an artificial rect region to replace the polygon region
"name": "rect",
"x": x,
"y": y,
"width": width,
"height": height
}
file["regions"].append(rectRegion) #append this newly created rect region to regions of the current file
elif(region["shape_attributes"]["name"] == "polyline"): #if the region has the shape of a polyline (Polylines cannot be handled in later steps so they have to be converted into rects)
linetemp = [];
for i in range(len(region["shape_attributes"]["all_points_x"])): #get all points of the polyline
coordtemp = [region["shape_attributes"]["all_points_x"][i], region["shape_attributes"]["all_points_y"][i]]
linetemp.append(coordtemp) #save them in a list
polyline = LineString(linetemp) #create a python LineString object out of them
line_bbox = polyline.bounds #get the bounding box (a list of the coordinates of the surrounding rectangle) of this newly created linestring/polyline
x = line_bbox[0] #save associated x value of bounding box
y = line_bbox[1] #save associated y value of bounding box
width = line_bbox[3] - line_bbox[0] #calculate width out of bounding box coordinates
height = line_bbox[3] - line_bbox[0] #calculate height out of bounding box coordinates
rectRegion = { #create an artificial rect region to replace the polyline region
"name": "rect",
"x": x,
"y": y,
"width": width,
"height": height
}
file["regions"].append(rectRegion) #append this newly created rect region to regions of the current file
else: #the region is of the type rect
file["regions"].append(region["shape_attributes"]) #just append the attributes of the rect regions to regions of the current file
croppedList = [] #list for the individual images of flower regions
for values in datasetDict_flower.values(): #loop through each dataset and their values
os.chdir("pathxy") #change directory to folder where images of each dataset is saved
os.chdir(pictureRelations[values["csv_Metadata"]]) #change directory to image folder of current dataset
for file in values["files"]: #loop through all image filenames of current dataset
image = Image.open(file["filename"]) #open current imagefile as python image
for region in file["regions"]: #loop through all region coordinates in current imagefile
#get coordinates
x = int(region["x"])
y= int(region["y"])
x2 = int(region["x"] + 150) #dirty solution to get the right image size see jsonReader_flowerVase_v2 for better solution
y2 = int(region["y"] + 200) #dirty solution to get the right image size see jsonReader_flowerVase_v2 for better solution
cropped = image.crop((x, y, x2, y2)) #crop image according to coordinates of current region
croppedList.append(cropped) #append this cropped image to list
os.chdir("pathxy") #change directory to folder where you want to save the newly created images of all flower regions
for i in range(len(croppedList)): #loop through all cropped images and save them as jpeg
filename = str(i) + "cropped_flower.jpeg"
croppedList[i].save(filename)