def pre_process_data(images):
"""Pre-process the data folder to find all images and crop all faces.
Return:
ImageSet of all faces.
"""
face_haar_cascade = HaarCascade('face2.xml')
cropped_images = ImageSet()
for image in images:
cropped_images.extend(get_faces(image, face_haar_cascade))
return cropped_images
def build_bg():
for n in range(BG_PICS):
pn = bgname.format(n)
save_picture(pn)
sleep(FRAME_DELAY)
camera.stop_preview()
frames = ImageSet()
frames.load(bgdir)
img = Image(frames[0].size())
nframes = len(frames)
for frame in frames:
img = img + (frame / nframes)
img.save(bgf)
return img
def run_trainer():
snack_trainer = SnackTrainer()
snack_trainer.parse_options(sys.argv)
debug = snack_trainer.options.debug
if snack_trainer.options.classes == "":
classes = snack_trainer.getClassNameFromPath(
snack_trainer.options.train_path)
else:
classes = snack_trainer.options.classes.split(',')
trainPaths = [snack_trainer.options.train_path + '/' + c for c in classes]
testPaths = [snack_trainer.options.test_path + '/' + c for c in classes]
extractors = [
snack_trainer.createExtractor('hue'),
snack_trainer.createExtractor('edge'),
snack_trainer.createExtractor('haar')
]
resultPath = snack_trainer.options.result_path
featurePath = snack_trainer.options.feature_path
classifierFile = snack_trainer.options.classifier_file
classifier_name = snack_trainer.options.classifier
classifier = snack_trainer.createClassifier(classifier_name, extractors)
snack_trainer.setClassifier(classifier)
print "Using Classifier:", classifier_name
print "Training Set:", trainPaths
print "Training Features Save As:", featurePath
snack_trainer.trainClassifier(classes, snack_trainer.options.train_path,
featurePath)
if classifierFile == "":
classifierFile = "static/trainer/" + "%s.dat" % (classifier_name)
print "Classifier Data Save As:", classifierFile
classifier.save(classifierFile)
if (debug):
print "Testing Set:", testPaths
imgs = ImageSet()
for p in testPaths:
imgs += ImageSet(p)
random.shuffle(imgs)
snack_trainer.testClassifier(classifier, classes, testPaths)
print "Test Result:", resultPath
snack_trainer.saveResults(classifier, imgs, resultPath)
print "Done"
def analyze(self):
image_set = ImageSet(self.in_path)
image_set.sort()
report_gen = CSVReportGenerator(os.path.join(self.report_path, 'noise_analysis.csv'))
counter = 1
print 'Extracting using %s' % self.method_name
for image in sorted(image_set, key=lambda img: img.filename):
filename = os.path.splitext(os.path.basename(image.filename))[0]
noise = self.analyze_single(image)
#print 'Analyzing %d/%d: %s' % (counter, len(image_set.filelist), filename)
print '%.2f' % noise
counter += 1
report_gen.generate()
def GetSpecificCardData(suit, rank, path="./data/", label=True):
fullpath = path + "/" + suit + "/" + rank + "/"
print fullpath
iset = ImageSet(fullpath)
if (label):
label_vals = []
l = (suit, rank) # not sure this is how we want to do this
sz = len(iset)
for i in range(0, sz):
label_vals.append(l)
return iset, label_vals
else:
return iset
def main():
data_batch_1 = ['/Users/linting/Desktop/CSC522/Project/simplecv/data/cifar-10-batches-mat/data_batch_1/'+c for c in classes ]
data_batch_2 = ['/Users/linting/Desktop/CSC522/Project/simplecv/data/cifar-10-batches-mat/data_batch_2/'+c for c in classes ]
data_batch_3 = ['/Users/linting/Desktop/CSC522/Project/simplecv/data/cifar-10-batches-mat/data_batch_3/'+c for c in classes ]
data_batch_4 = ['/Users/linting/Desktop/CSC522/Project/simplecv/data/cifar-10-batches-mat/data_batch_4/'+c for c in classes ]
data_batch_5 = ['/Users/linting/Desktop/CSC522/Project/simplecv/data/cifar-10-batches-mat/data_batch_5/'+c for c in classes ]
trainPaths = data_batch_1 + data_batch_2 + data_batch_3 + data_batch_4 + data_batch_5
testPaths = ['/Users/linting/Desktop/CSC522/Project/simplecv/data/cifar-10-batches-mat/test_batch/'+c for c in classes]
trainer = Trainer(classes,trainPaths)
print "Start training....."
trainer.train()
imgs = ImageSet()
for p in testPaths:
imgs += ImageSet(p)
random.shuffle(imgs)
print "Result test"
trainer.test(testPaths)
def extract(self):
image_set = ImageSet(self.data_path)
image_set.sort()
output_path = os.path.join(self.result_path,
self.__experiment.name + '.csv')
report_gen = CSVReportGenerator(output_path, params=self.params)
counter = 1
print 'Extracting using %s' % self.method_name
for image in sorted(image_set, key=lambda img: img.filename):
filename = os.path.splitext(os.path.basename(image.filename))[0]
instance = session.query(Analysis).filter_by(
filename=image.filename,
experiment_name=self.__experiment.name,
experiment_params=self.__experiment.params).first()
if instance:
analysis = instance
else:
try:
analysis = self.analyze_mono_bead(image)
analysis.experiment = self.__experiment
session.add(analysis)
print 'Analyzing %d/%d: %s' % (
counter, len(image_set.filelist), filename)
except NoBeadException:
print 'No Bead Exception: %s' % filename
# Sholl Analysis Plots
#self.plot_sholl_analysis(analysis, filename)
# Add to overall report
report_gen.add_analysis(filename, analysis)
counter += 1
session.commit()
report_gen.generate()
def GetFullDataSet(path="./data", label=True): # just load everything
data = ImageSet()
labels = []
for s in SUITS:
for r in RANKS:
if (label):
d, l = GetSpecificCardData(s, r, path, label)
data += d
labels += l
else:
d = GetSpecificCardData(s, r, path, label)
data += d
if (label):
return data, labels
else:
return data
def train(self, digits_path):
for digit_to_train in range(-1, 10):
logging.info("Training digit " + str(digit_to_train))
digit_images = ImageSet(digits_path + str(digit_to_train))
logging.debug("Loaded " + str(len(digit_images)) + " digits")
for digit_image in digit_images:
self.train_image(digit_image, digit_to_train)
logging.info("Digit " + str(digit_to_train) + " trained")
self.train_network()
#!/usr/bin/env python
import logging
logging.getLogger().setLevel(logging.INFO)
from SimpleCV import ImageSet
from GasMeter import GasMeter
def is_number(s):
try:
float(s)
return True
except ValueError:
return False
test_frames = ImageSet("./images/input")
for frame in test_frames:
gas = GasMeter(frame)
value = gas.get_meter_value()
if is_number(value):
print(frame.filename + "\t" + "{0:.3f}".format(float(value)))
else:
print(frame.filename + "\t" + value + " UNRELIABLE")
import os
from time import sleep
from SimpleCV import ImageSet
from utils import PreparedCamera
# Settings
cam_size = {'width': 640, 'height': 480}
gif_interval = 0.5 # Seconds between frames
gif_length = 10 # Length in seconds
file_name = os.path.join('pictures', 'ani.gif')
cam = PreparedCamera(0, cam_size, debug=True)
print("Recording images")
img_set = ImageSet()
for frame in range(0, int(gif_length * gif_interval)):
img = cam.getImage()
img_set.append(img)
sleep(1.0/gif_interval)
img_set.save(destination=file_name, dt=gif_interval)
from SimpleCV import ImageSet
set = ImageSet(".")
for img in set:
oldname = img.filename
newname = oldname[0:-3] + 'jpg'
print "Converting " + oldname + " to " + newname
img.save(newname)
def main( args ):
#
# Set these flags so they can be checked for validity later.
#
res = None
fac = None
#
# Check for any args that can safely be defaulted.
#
if args.dest:
outdir = args.dest
#
# Append a trailing slash if it's not there.
#
if outdir[-1] != '/':
outdir = outdir + '/'
else:
outdir = 'scaled_thumbs/'
#
# Check for scale or resolution. The default action is to scale
# by 1/2.
#
if args.res:
(width, height) = args.res.split('x')
#
# Just set this to something other than None, so the script can
# tell whether or not to use the resolution later.
#
res = args.res
elif args.fac:
#
# Limit the scale factor so super-large images are made even bigger.
#
fac = float(args.fac)
if fac >= 2.0:
fac = 2.0
if fac <= 0.0:
fac = 0.1
else:
fac = .5
#
# Make sure the new files have a home.
#
if not isdir(outdir):
mkdir(outdir)
#
# Load the files from the command line
#
images = ImageSet()
_info_('Adding images to image set... this may take a moment.')
for item in args.files:
#
# Make sure I don't try to convert something that's not an image.
#
try:
image = Image(item)
except IOError, err:
_err_("Invalid/unuseable file: " + item)
exit(2)
images.append(image)
from SimpleCV import Image, Display, ImageSet, Color, ROI
import numpy as np
import os as os
sz = 256
count = 0
outdir = './faces/'
if not os.path.exists(outdir):
os.mkdir(outdir)
negTotal = 0
negcount = 5
negdir = './negs/'
if not os.path.exists(negdir):
os.mkdir(negdir)
disp = Display()
iset = ImageSet('data')
for img in iset:
img.save(disp)
face = img.findHaarFeatures('face.xml')
if (face is not None and len(face) > 0):
face[-1].draw(width=3)
img.applyLayers().save(disp)
fname = outdir + 'Face' + str(count) + '.png'
face[-1].crop().resize(sz, sz).toGray().save(fname)
print "Saving: " + fname
count = count + 1
ncount = 0
while ncount < negcount:
pt = 256 * np.random.rand(1, 2)
roi = ROI(
np.round((img.width - sz) * np.random.rand(1))[0],
from SimpleCV import Camera, ImageSet
import time
cam = Camera()
camImages = ImageSet()
# Set to a maximum of 10 images saved
# Feel free to increase, but bewared running out of space
maxImages = 10
for counter in range(maxImages):
# Capture a new image and add to set
img = cam.getImage()
camImages.append(img)
# Show the image and wait before capturing another
img.show()
time.sleep(6)
camImages.save(verbose=True)
def __init__(self,
source_paths,
out_path,
classList,
key_bindings,
preprocess=None,
postprocess=None):
#if( not os.access(out_path,os.W_OK) ):
# print "Output path is not writeable."
# raise Exception("Output path is not writeable.")
self.keyBindings = key_bindings
self.classes = classList
self.countMap = {}
self.classMap = {}
self.directoryMap = {}
self.out_path = out_path
self.keyMap = {}
if (len(classList) != len(key_bindings)):
print "Must have a key for each class."
raise Exception("Must have a key for each class.")
for key, cls in zip(key_bindings, classList):
self.keyMap[key] = cls
# this should work
if (preprocess is None):
def fakeProcess(img):
return [img]
preprocess = fakeProcess
self.preProcess = preprocess
if (postprocess is None):
def fakePostProcess(img):
return img
postprocess = fakePostProcess
self.postProcess = postprocess
self.srcImgs = ImageSet()
if (isinstance(source_paths, ImageSet)):
self.srcImgs = source_path
else:
for sp in source_paths:
print "Loading " + sp
imgSet = ImageSet(sp)
print "Loaded " + str(len(imgSet))
self.srcImgs += imgSet
if (not osp.exists(out_path)):
os.mkdir(out_path)
for c in classList:
outdir = out_path + c + '/'
self.directoryMap[c] = outdir
if (not osp.exists(outdir)):
os.mkdir(outdir)
for c in classList:
searchstr = self.directoryMap[c] + '*.png'
fc = glob.glob(searchstr)
self.countMap[c] = len(fc)
self.classMap[c] = ImageSet(self.directoryMap[c])
from SimpleCV import Camera, ImageSet
import time
cam = Camera()
camImages = ImageSet()
maxImages = 10
for i in range(maxImages):
# Capture a new image
img = cam.getImage()
# Add image to set
camImages.append(img)
# Show the image and wait before capturing another
img.show()
time.sleep(1)
camImages.save(destination="./images/imagesets/")
import numpy
from TemplateDigitDetector import TemplateDigitDetector
from SVCDigitDetector import SVCDigitDetector
# Simple script for evaluating Digit detectors
svc_detector = SVCDigitDetector()
template_detector = TemplateDigitDetector()
svc_detector_success = []
template_detector_success = []
for test_digit in range(10):
test_samples = ImageSet("./images/dataset-test/" + str(test_digit))
for test_sample in test_samples:
svc_digit = svc_detector.detect_digit(test_sample)
template_digit = template_detector.detect_digit(test_sample)
print("svc_digit:" + str(svc_digit) + " template_digit:" +
str(template_digit))
if svc_digit != test_digit:
svc_detector_success.append(0)
else:
svc_detector_success.append(100)
if template_digit != test_digit:
from SimpleCV import ImageSet
import time
images = ImageSet("C:\Users\saurav datta\Documents\python learning\simpleCV learnings")
images.show(2) # 2nd argument is delay between slide show
# display all the images ina folder as slide show
from SimpleCV import Camera, ImageSet import time cam = Camera() camImages = ImageSet() # Set to a maximum of 10 images saved # Feel free to increase, but beware of running out of space maxImages = 5 for counter in range(maxImages): # Capture a new image and add to set img = cam.getImage() camImages.append(img) # Show the image and wait before capturing another img.show() time.sleep(2) camImages.save(destination='imageset1', verbose=True)
from SimpleCV import ImageSet
img_set = ImageSet(".")
for img in img_set:
oldname = img.filename
newname = oldname[0:-3] + 'jpg'
print 'Converting ' + oldname + ' to ' + newname
img.save(newname)
y2 = np.round(imgsz * np.random.rand(1, nDesc))[0].tolist()
p1 = zip(x1, y1)
p2 = zip(x2, y2)
descriptors = zip(p1, p2)
def extract(iset, nDesc, descriptors):
posCounts = np.zeros([1, nDesc])[0]
for face in iset:
for idx, d in enumerate(descriptors):
if (face[d[0][0], d[0][1]][0] > face[d[1][0], d[1][1]][0]):
posCounts[idx] += 1
return np.array(posCounts) / len(iset)
faces = ImageSet('./faces/')
faceResult = extract(faces, nDesc, descriptors)
print faceResult
notFaces = ImageSet('./negs/')
negResult = extract(notFaces, nDesc, descriptors)
print negResult
fig = plt.figure()
ax = fig.add_subplot(111)
ind = np.arange(nDesc)
width = 0.5
rects1 = ax.bar(ind, faceResult, width, color='r')
rects2 = ax.bar(ind + width, negResult, width, color='b')
# add some
ax.set_ylabel('Percentage with descriptor == 1')
ax.set_title('Random Binary Descriptor Incidence')
ax.legend((rects1[0], rects2[0]), ('faces', 'notfaces'))
from SimpleCV import ColorSegmentation, Image, Display, ImageSet
import time
redBlock = Image('redblock.png')
greenBlock = Image('greenblock.png')
blueBlock = Image('blueblock.png')
cs = ColorSegmentation()
cs.addToModel(redBlock)
cs.addToModel(greenBlock)
cs.addToModel(blueBlock)
cards = ImageSet('cards')
card = None
disp = Display((320, 240))
score = 0
isPrimary = False
while (cards or card) and disp.isNotDone():
if card is None:
card = cards.pop()
cs.addImage(card)
res = cs.getSegmentedImage()
color = res.meanColor()
if ((color[0] < 254) and (color[1] < 254) and (color[2] < 254)):
isPrimary = True
# -*- coding: utf-8 -*-
from SimpleCV import ImageSet
set = ImageSet("ImageSetFolder/.")
for img in set:
oldname = img.filename
newname = oldname[0:-3] + 'jpg'
print "Converting " + oldname + " to " + newname
img.save(newname)
