posdatadir= ""

negdatadir= ""

trainfiles = []

boundaryboxes = []

imageseriesname = ""

trainboxpatches = []

bbfilenames = []

negpatches = [] # negative training examples

def __init__(self):

config = cp.RawConfigParser()

config.read('config.cfg')

self.datadir = config.get('init', 'posdatadir')

self.negdatadir = config.get('init', 'negdatadir')

# Set the image series name

temp = self.datadir.split('/')[-1]

temp = temp.strip(" \r\n\"")

self.imageseriesname = temp

# Get the traiing and boundary boxes

self.gen_train_files()

self.get_boundary_boxes()

self.get_trainbox_patches()

self.gen_neg_examples()

def rgb2gray(self, rgb):

img = color.rgb2gray(rgb);

# r, g, b = rgb[:,:,0], rgb[:,:,1], rgb[:,:,2]

# gray = 0.2989 * r + 0.5870 * g + 0.1140 * b

return img

def gen_train_files(self):

"""

Gets the training data file

"""

files = os.listdir(self.datadir)

self.imgfiles = [x for x in files if 'jpg' or 'png' or 'bmp' in files]

def get_boundary_boxes(self):

"""

Reads the boundary box files

"""

files = os.listdir(self.datadir)

bbfiles = [x for x in files if 'groundtruth' in x]

# Get the bounding boxes

allboxes = []

bbfilenames = []

for b in bbfiles:

s = open(self.datadir +'/' +b).read()

boxes = s.split('\n')

boxes = [x.split(' ')[ :4] for x in boxes]

boxes = [x for x in boxes if x !=['']]

allboxes += boxes

# Adding file names corresponding to boundary boxes

fname = b.replace('_'+self.imageseriesname.lower() + '.groundtruth', '.jpg')

for i in range(len(boxes)):

bbfilenames.append(fname)

# Setting boundary boxes and associated filenames

self.boundaryboxes = allboxes

self.bbfilenames = bbfilenames

def get_trainbox_patches(self):

if self.boundaryboxes == []:

raise ' Boundary box is empty. call get_boundary_boxes to fill in this array!\n'

# For each bounding box read the image and generate the boxes

for i in range(len(self.bbfilenames)):

fname = self.bbfilenames[i]

bbox = self.boundaryboxes[i]

x1 = float(bbox[0])

y1 = float(bbox[1])

x2 = float(bbox[2])

y2 = float(bbox[3])

# read the image

im = cv2.imread(self.datadir + '/' + fname)

#gray = self.rgb2gray(im)

gray = cv2.cvtColor(im, cv2.COLOR_BGR2GRAY)

# resize the trainboxpatches

patch = gray[y1:y2, x1:x2]

patch = cv2.resize(patch, (64,64))

self.trainboxpatches.append(gray[y1:y2, x1:x2])

def gen_neg_examples(self):

"""

Generates negative training patches

from the dataset specified in the configuration

"""

negfiles = os.listdir(self.negdatadir)

negpatches = []

for n in negfiles[:10]:

f = self.negdatadir + '/' + n

if 'jpg' not in f:

continue

print 'Processing image'+ f

im = cv2.imread(f)

#gray = self.rgb2gray(im)

gray = cv2.cvtColor(im, cv2.COLOR_BGR2GRAY)

#im =cv2.cvtColor(im,cv2.COLOR_BGR2GRAY)

# Generate the negative patch

#patch = image.extract_patches_2d(im, (256, 256)) # This generalates nultiple patches

print np.shape(im)

patch = view_as_windows(gray, (64,64),65)

print np.shape(patch)

#patch = [cv2.resize(x,(64,64)) for x in patch]

# randomly choose 20 patches from all patches for this image

fdimlen = np.shape(patch)[0]

sdimlen = np.shape(patch)[1]

patches = []

#for i in range(len(self.trainpatches)):

for i in range(8): # we are getting too many negative examples, so reducing to 5

randomdim = random.choice(range(fdimlen ))

randomsdim = random.choice(range(sdimlen))

#randompatch = patch[randomdim][randomsdim][0] # if color un comment this

randompatch = patch[randomdim][randomsdim]

#randompatch = cv2.resize(randompatch, (64,64))

patches.append(randompatch)

negpatches += patches