def main_fun(self):
# initialize the color descriptor
cd = ColorDescriptor((8, 12, 3))
# open the output index file for writing
output = open(self.path, "w")
#data contains the path to your images folder
data = '/home/farhan/project/CBIR/my_contrib/test_dataset'
# use glob to grab the image paths and loop over them
types = ['/*.png', '/*.jpg']
# os.chdir(data)
for filetype in types:
for imagePath in glob.glob(data + filetype):
# extract the image ID (i.e. the unique filename) from the image
# path and load the image itself
imageID = imagePath[imagePath.rfind("/") + 1:]
image = cv2.imread(imagePath)
# describe the image
features = cd.describe(image)
# write the features to file
features = [str(f) for f in features]
output.write("%s,%s\n" % (imageID, ",".join(features)))
# close the index file
output.close()
def main_fun(self):
# initialize the color descriptor
cd = ColorDescriptor((8, 12, 3))
# open the output index file for writing
output = open(self.path, "w")
#data contains the path to your images folder
data='/home/farhan/project/CBIR/my_contrib/test_dataset'
# use glob to grab the image paths and loop over them
types=['/*.png','/*.jpg']
# os.chdir(data)
for filetype in types:
for imagePath in glob.glob(data+filetype):
# extract the image ID (i.e. the unique filename) from the image
# path and load the image itself
imageID = imagePath[imagePath.rfind("/") + 1:]
image = cv2.imread(imagePath)
# describe the image
features = cd.describe(image)
# write the features to file
features = [str(f) for f in features]
output.write("%s,%s\n" % (imageID, ",".join(features)))
# close the index file
output.close()
def searchimg(self, index1='index1.csv', index2='index2.csv', result_path='jpg'):
cd = ColorDescriptor((8,12,3))
vd = Vgg16Descriptor((224,224,3))
query = cv2.imread("./query/" + self.filename)
feature1 = cd.describe(query)
feature2 = vd.describe(self.querypath)
searcher = Searcher(index1,index2)
results = searcher.search(feature1,feature2)
result0 = cv2.resize(query,(128,128),interpolation=cv2.INTER_CUBIC)
name = locals()
i = 1
for (score, resultID) in results:
name['result%d'%i] = cv2.imread(result_path + "/" + resultID)
name['result%d'%i] = cv2.resize(name['result%d'%i],(128,128),interpolation=cv2.INTER_CUBIC)
i = i + 1
result_0 = np.hstack(name['result%d'%i] for i in range(1,6))
result_1 = np.hstack(name['result%d'%i] for i in range(6,11))
result = np.vstack((result_0,result_1))
cv2.imwrite("./result/%s" % self.filename, result)
return self.filename
def indexpy(dataset, index):
print('hello')
cd = ColorDescriptor((8, 12, 3))
output = open(index, "w")
for imagePath in glob.glob(dataset + "/*.jpg"):
imageId = imagePath[imagePath.rfind("/") + 1:]
image = cv2.imread(imagePath)
features = cd.describe(image)
features = [str(f) for f in features]
output.write("%s,%s\n" % (imageId, ",".join(features)))
output.close()
def search(img):
# initialize the image descriptor
cd = ColorDescriptor((8, 12, 3))
# load the query image and describe it
#query = cv2.imread(args["query"])
#query = cv2.imread("queries/2.png")
features = cd.describe(img)
# perform the search
searcher = Searcher("feature.csv")
results = searcher.search(features)
return results
def search_cut(self):
# initialize the image descriptor
cd = ColorDescriptor((8, 12, 3))
# load the query image and describe it
query = cv2.imread("out.jpg")
features = cd.describe(query)
# perform the search
searcher = Searcher("index.csv")
results = searcher.search(features)
# loop over the results
for (score, resultID) in results:
# load the result image and display it
result = cv2.imread(resultID)
cv2.imshow("Result", result)
cv2.waitKey(0)
def create_features(path):
# initialize the color descriptor
cd = ColorDescriptor((4, 4, 2))
# use glob to grab the image paths and loop over them
image_names, X = [], []
for imagePath in listdir(path):
# extract the image ID (i.e. the unique filename) from the image
# path and load the image itself
imageID = basename(imagePath).split('.')[0]
image = cv2.imread(join(path, imagePath))
# describe the image
features = cd.describe(image)
image_names.append(imageID)
X.append(features)
return np.array(image_names), np.array(X)
def create_features(path):
# initialize the color descriptor
cd = ColorDescriptor((4, 4, 2))
# use glob to grab the image paths and loop over them
image_names, X = [], []
for imagePath in listdir(path):
# extract the image ID (i.e. the unique filename) from the image
# path and load the image itself
imageID = basename(imagePath).split('.')[0]
image = cv2.imread(join(path,imagePath))
# describe the image
features = cd.describe(image)
image_names.append(imageID)
X.append(features)
return np.array(image_names), np.array(X)
def search(index, query, result_path):
# initialize the image descriptor
cd = ColorDescriptor((8, 12, 3))
# load the query image and describe it
query1 = cv2.imread(query)
features = cd.describe(query1)
# perform the search
searcher = Searcher(index)
results = searcher.search(features)
# display the query
#cv2.imshow("Query", query1)
search_output = []
for (score, resultID) in results:
#print(result_path+"/"+resultID)
full_path = result_path + "/" + resultID
#result=cv2.imread(result_path+"/"+resultID)
#cv2.imshow("Result", result)
#cv2.waitKey(0)
search_output.append(resultID)
#print(search_output)
return search_output
def upload_file():
if request.method == 'POST':
# check if the post request has the file part
if 'file' not in request.files:
flash('No file part')
return redirect(request.url)
file = request.files['file']
if file.filename == '':
flash('No file selected for uploading')
return redirect(request.url)
if file and allowed_file(file.filename):
filename = secure_filename(file.filename)
file.save(os.path.join(app.config['UPLOAD_FOLDER'], filename))
# initialize the image descriptor
cd = ColorDescriptor((8, 12, 3))
# load the query image and describe it
photos_result = []
query = cv2.imread(app.config['UPLOAD_FOLDER']+"/"+filename)
features = cd.describe(query)
query_photo = {"score": 100, "photo": "queries/"+filename}
photos_result.append(query_photo)
# perform the search
searcher = Searcher("./index.csv")
results = searcher.search(features)
# loop over the results
for (score, resultID) in results:
score = log10(abs(score))
# load the result image and display it
result = {"score": score, "photo": "dataset/"+resultID}
photos_result.append(result)
return render_template('result.html', results=photos_result)
else:
flash('Allowed file types are txt, pdf, png, jpg, jpeg, gif')
return redirect(request.url)
def search(filename):
#print(query)
cd = ColorDescriptor((8, 12, 3))
query = cv2.imread(filename)
query = cv2.resize(query, (416, 416))
filepath = 'static/' + filename
cv2.imwrite(filepath, query)
features = cd.describe(query)
searcher = Searcher()
results = searcher.search(features)
#cv2.imshow("Query", query)
# loop over the results
for (score, resultID) in results:
# load the result image and display it
#print(resultID)
img = mongo.db.myImages.find({'filename': resultID}, {
"_id": 0,
"name": 0,
"filename": 0
})
l = []
for x in img:
#print(x)
x = x['images'][0]
#print('hey')
#print(x)
fout = fs.get(x['imageID'])
im = np.frombuffer(fout.read(), dtype=np.uint8)
im = np.reshape(im, x['shape'])
# cv2.imshow("Result", im)
# cv2.waitKey(0)
return results
def main_search(self):
cd = ColorDescriptor((8, 12, 3))
# load the query image and describe it
query = cv2.imread('test_dataset/' + self.name)
features = cd.describe(query)
# perform the search
searcher = Searcher(self.file_path)
results = searcher.search(features)
# display the query
cv2.imshow("Query", query)
# loop over the results
for (score, resultID) in results:
# load the result image and display it
result = cv2.imread(self.data_path + "/" + resultID)
r = 300.0 / result.shape[1]
dim = (300, int(result.shape[0] * r))
resized = cv2.resize(result, dim, interpolation=cv2.INTER_AREA)
cv2.imshow("RESULT", resized)
cv2.waitKey(0)
cv2.destroyAllWindows()
def main_search(self):
cd = ColorDescriptor((8, 12, 3))
# load the query image and describe it
query = cv2.imread('test_dataset/'+self.name)
features = cd.describe(query)
# perform the search
searcher = Searcher(self.file_path)
results = searcher.search(features)
# display the query
cv2.imshow("Query",query)
# loop over the results
for (score, resultID) in results:
# load the result image and display it
result = cv2.imread(self.data_path + "/" + resultID)
r=300.0/result.shape[1]
dim=(300,int(result.shape[0]*r))
resized = cv2.resize(result, dim, interpolation = cv2.INTER_AREA)
cv2.imshow("RESULT", resized)
cv2.waitKey(0)
cv2.destroyAllWindows()
fea = []
final_fea = {}
output = open(args["index"], "w")
# use glob to grab the image paths and loop over them
for imagePath in glob.glob(args["dataset"] + "/*.jpg"):
# extract the image ID (i.e. the unique filename) from the image
# path and load the image itself
imageID = imagePath[imagePath.rfind("/") + 1:]
print imagePath, "imagePath"
# print imageID,"imageID"
image = cv2.imread(imagePath)
# print image[210,210,1],"pixel"
#print imageID,"imageID"
# describe the image
features = cd.describe(image)
# write the features to file
#features = [str(f) for f in features]
#output.write("%s\n" % (imageID))
#output.write("%s\n" % (imageID))
#for f in features:
#fea.append(f)
#print f,"fea"
final_fea[imageID] = features
print len(final_fea), "shape"
# print final_fea,"final_fea"
# load the query image and describe it
output.write("%s;" % ("testing image name"))
output.write("%s\n" % ("training image name"))
for imagePath1 in glob.glob(args["query"] + "/*.jpg"):
imageID1 = imagePath1[imagePath1.rfind("/") + 1:]
ap.add_argument(
"-d",
"--dataset",
required=True,
help="Path to the directory that contains the images to be indexed")
ap.add_argument("-i",
"--index",
required=True,
help="Path to where the computed index will be stored")
args = vars(ap.parse_args())
cd = ColorDescriptor((8, 12, 3))
output = open(args["index"], "w")
# use glob to grab the image paths and loop over them
for imagePath in glob.glob(args["dataset"] + "/*.png"):
imageID = imagePath[imagePath.rfind("/") + 1:]
image = cv2.imread(imagePath)
# describe the image
features = cd.describe(image)
# write the features to file
features = [str(f) for f in features]
output.write("%s,%s\n" % (imageID, ",".join(features)))
# close the index file
output.close()
from colordescriptor import ColorDescriptor
from searcher import Searcher
import argparse
import cv2
ap = argparse.ArgumentParser()
ap.add_argument("-i", "--index", required=True, help="Path to where the computed index will be stored")
ap.add_argument("-q", "--query", required=True, help="Path to the query image")
ap.add_argument("-r", "--result-path", required=True, help="Path to the result path")
args = vars(ap.parse_args())
cd = ColorDescriptor((8, 12, 3))
query = cv2.imread(args["query"])
features = cd.describe(query)
searcher = Searcher(args["index"])
results = searcher.search(features)
query = cv2.resize(query, (960, 540))
cv2.namedWindow("Query", cv2.WINDOW_NORMAL)
cv2.imshow("Query", query)
cv2.waitKey(0)
for (score, resultID) in results:
result = cv2.imread(args["result_path"] + "/" + resultID)
#result = cv2.resize(result, (960, 540))
cv2.namedWindow("Result", cv2.WINDOW_NORMAL)
cv2.imshow("Result", result)
#print(args["result_path"] + "/" + resultID)
cv2.waitKey(0)
# initialize the color descriptor
cd = ColorDescriptor(
(8, 12, 3)) # HSV binning -- 8 hue, 12 saturation, 3 value
features = []
ctrcorr = []
images = glob.glob(folderPath + '/*.jpg')
for pic in images:
image = cv2.imread(pic)
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
ctr = []
temp = cd.describe(image)
features.append(temp)
for i in temp[0:4]:
ctr.append(np.corrcoef(i, temp[4])[1, 0])
ctrcorr.append(ctr)
names = [t[t.rfind("/") + 1:] for t in images]
zippy = zip(features, names, ctrcorr)
""" Perform 3 kinds of checks:
1) How many peaks are there within a certain range of the first peak? Uniform background will have relatively few high peaks while variable background will have many lower peaks and no strong signals. Measure by counting the number of peaks within some threshold below the highest peak [peak-thresh:peak].
1a) How dissimilar are the corners from the center of the image? If the corners are extremely uniform, this is a fairly lenient check (correlation between corners and center < 0.998, mostly just gets rid of tiled images), since often a fish will be partially transparent or similar in color to the background.
2) Of those without a single strong peak, are the four corners extremely similar? Measure using correlation coefficients between four corners. If at least two are above the threshold then we have at least two pairs of corners that are extremely similar (usually top two and bottom two). If # (corrcoef >= failCorr) >= 2 then keep image.
2a) If the corners are fairly similar but not extremely similar, check whether the center of the image is extremely dissimilar from the corners (meaning there's probably a fish taking up most of the center).
Take images that passed one of two checks and copy them to a 'Pass' folder within current WD; put remaining images in a 'Fail' folder - can be accessed later to check for any missing images.
import argparse
import cv2
import time
# construct the argument parser and parse the arguments
ap = argparse.ArgumentParser()
ap.add_argument("-i", "--index", default='index.csv',
help="Path to where the computed index will be stored")
ap.add_argument("-l", "--limit", default=10,
help="Query result limit")
ap.add_argument("-d", "--dist", default='euclidean',
help="Distance model")
ap.add_argument("-q", "--query", required=True,
help="Path to the query image")
args = vars(ap.parse_args())
# initialize the image descriptor
cd = ColorDescriptor((8, 12, 3))
# load the query image and describe it
query = cv2.imread(args["query"])
features = cd.describe(query)
# perform the search
searcher = Searcher(args["index"])
results = searcher.search(features, int(args["limit"]), args["dist"])
# loop over the results
for (score, resultID) in results:
# load the result image and display it
print resultID
import cv2
# construct an argument parser
ap = argparse.ArgumentParser()
ap.add_argument("-d", "--dataset", required=True,
help = "Path to the directory that contains the images to be indexed")
ap.add_argument("-i", "--index", required=True,
help = "Path of file where the computed index will be stored")
# vars() return a dict, it can be retrieved as args[dataset]/args[index]
args = vars(ap.parse_args())
# in HSV, 8 bins for hue, 12 bins for saturation, 3 for values(brightness)
cd = ColorDescriptor((8, 12, 3))
# open output index file for writing
output = open(args["index"], "w")
# use glob to grab the image files in path and iterate them
for imagePath in glob.glob(args["dataset"] + "/*.jpg"):
# extract the image ID(i.e. the unique filename)
imageID = imagePath[imagePath.rfind('/') + 1 :]
image = cv2.imread(imagePath)
features = cd.describe(image)
# write the features to file
features = [str(f) for f in features]
output.write("%s,%s\n" % (imageID, ",".join(features)))
output.close()
import glob
import cv2
# Parse arguments
argParser = argparse.ArgumentParser()
argParser.add_argument("-d", "--dataset", required=True,
help="Path to directory that contains the images to be indexed")
argParser.add_argument("-i", "--index", required=True,
help="Path to where teh computed idnex will be stored")
args = vars(argParser.parse_args())
# 8 hue bins, 12 saturation bins, 3 value bins
colorDescriptor = ColorDescriptor((8, 12, 3))
# Open the indexFile in which we will save the indexed images
indexFile = open(args["index"], "w")
# Loop over image files with glob
for imagePath in glob.glob(args["dataset"] + "/*.jpg"):
imageID = imagePath[imagePath.rfind("\\") + 1:]
image = cv2.imread(imagePath)
# get the features
features = colorDescriptor.describe(image)
# write feature to file
features = [str(f) for f in features]
indexFile.write("%s,%s\n" % (imageID, ",".join(features)))
indexFile.close()
ap.add_argument("-v",
"--index2",
required=True,
help="Path to where the vd computed index will be stored")
ap.add_argument("-q", "--query", required=True, help="Path to the query image")
ap.add_argument("-r",
"--result_path",
required=True,
help="Path to the result path")
args = vars(ap.parse_args())
cd = ColorDescriptor((8, 12, 3))
vd = Vgg16Descriptor((224, 224, 3))
query = cv2.imread(args['query'])
feature1 = cd.describe(query)
feature2 = vd.describe(args['query'])
searcher = Searcher(args["index1"], args["index2"])
results = searcher.search(feature1, feature2)
result0 = cv2.resize(query, (128, 128), interpolation=cv2.INTER_CUBIC)
name = locals()
i = 1
for (score, resultID) in results:
name['result%d' % i] = cv2.imread(args["result_path"] + "/" + resultID)
name['result%d' % i] = cv2.resize(name['result%d' % i], (128, 128),
interpolation=cv2.INTER_CUBIC)
i = i + 1
os.makedirs(folderPath + '/Fail')
# initialize the color descriptor
cd = ColorDescriptor((8, 12, 3)) # HSV binning -- 8 hue, 12 saturation, 3 value
features = []
ctrcorr = []
images = glob.glob(folderPath+'/*.jpg')
for pic in images:
image = cv2.imread(pic)
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
ctr = []
temp = cd.describe(image)
features.append(temp)
for i in temp[0:4]:
ctr.append(np.corrcoef(i, temp[4])[1,0])
ctrcorr.append(ctr)
names = [t[t.rfind("/")+1:] for t in images]
zippy = zip(features, names, ctrcorr)
""" Perform 3 kinds of checks:
1) How many peaks are there within a certain range of the first peak? Uniform background will have relatively few high peaks while variable background will have many lower peaks and no strong signals. Measure by counting the number of peaks within some threshold below the highest peak [peak-thresh:peak].
1a) How dissimilar are the corners from the center of the image? If the corners are extremely uniform, this is a fairly lenient check (correlation between corners and center < 0.998, mostly just gets rid of tiled images), since often a fish will be partially transparent or similar in color to the background.
2) Of those without a single strong peak, are the four corners extremely similar? Measure using correlation coefficients between four corners. If at least two are above the threshold then we have at least two pairs of corners that are extremely similar (usually top two and bottom two). If # (corrcoef >= failCorr) >= 2 then keep image.
# construct the argument parser and parse the arguments
ap = argparse.ArgumentParser()
ap.add_argument("-d", "--dataset", required = True,
help = "Path to the directory that contains the images to be indexed")
ap.add_argument("-i", "--index", required = True,
help = "Path to where the computed index will be stored")
args = vars(ap.parse_args())
# initialize the Shape descriptor
cd = ColorDescriptor(21)
# open the output index file for writing
output = open(args["index"], "w")
# use glob to grab the image paths and loop over them
for imagePath in glob.glob(args["dataset"] + "/*.png"):
# extract the image ID (i.e. the unique filename) from the image
# path and load the image itself
imageID = imagePath[imagePath.rfind("/") + 1:]
image = cv2.imread(imagePath)
# describe the image
features = cd.describe()
# write the features to file
features = [str(f) for f in features]
output.write("%s,%s\n" % (imageID, ",".join(features)))
# close the index file
output.close()
# -i = path to the csv index containing all feature vectors
from colordescriptor import ColorDescriptor
from searcher import Searcher
from imutils import build_montages
import argparse
import cv2
ap = argparse.ArgumentParser()
ap.add_argument("-i", "--index", type=str, default="index.csv")
ap.add_argument("-u", "--upload", required=True)
args = vars(ap.parse_args())
cd = ColorDescriptor((8, 12, 3))
upload = cv2.imread(args["upload"])
features = cd.describe(upload)
searcher = Searcher(args["index"])
results = searcher.search(features)
i = 0
locs = []
similars = []
for (score, resultID) in results:
result = cv2.imread(resultID)
if (i < 1):
if score < 1.0:
title = "Found: " + resultID
else:
title = "Not Found - Best Match: " + resultID
im = cv2.resize(result, (600, 400))
cv2.imshow(title, im)
