运行时报错：firstFolder = sys.argv[1]
IndexError: list index out of range
怎么回事？
import numpy as np
import cv2
import sys
from matplotlib import pyplot as plt
# img = cv2.imread('logo.png',0)
# # Initiate ORB detector
# orb = cv2.ORB_create()
# # find the keypoints with ORB
# kp = orb.detect(img,None)
# # compute the descriptors with ORB
# kp, des = orb.compute(img, kp)
# # draw only keypoints location,not size and orientation
# img2 = cv2.drawKeypoints(img, kp, None, color=(0,255,0), flags=0)
# plt.imshow(img2), plt.show()
from os import listdir
from os.path import isfile, join
class Application:
def __init__(self, extractor, detector):
self.extractor = extractor
self.detector = detector
def train_vocabulary(self, file_list, vocabulary_size):
kmeans_trainer = cv2.BOWKMeansTrainer(vocabulary_size)
for path_to_image in file_list:
img = cv2.imread(path_to_image, 0)
kp, des = self.detector.detectAndCompute(img, None)
kmeans_trainer.add(des)
return kmeans_trainer.cluster()
def extract_features_from_image(self, file_name):
image = cv2.imread(file_name)
return self.extractor.compute(image, self.detector.detect(image))
def extract_train_data(self, file_list, category):
train_data, train_responses = [], []
for path_to_file in file_list:
train_data.extend(self.extract_features_from_image(path_to_file))
train_responses.append(category)
return train_data, train_responses
def train_classifier(self, data, responses):
n_trees = 200
max_depth = 10
model = cv2.ml.RTrees_create()
eps = 1
criteria = (cv2.TERM_CRITERIA_MAX_ITER, n_trees, eps)
model.setTermCriteria(criteria)
model.setMaxDepth(max_depth)
model.train(np.array(data), cv2.ml.ROW_SAMPLE, np.array(responses))
return model
def predict(self, file_name):
features = self.extract_features_from_image(file_name)
return self.classifier.predict(features)[0]
def train(self, files_array, vocabulary_size=12):
all_categories = []
for category in files_array:
all_categories += category
vocabulary = self.train_vocabulary(all_categories, vocabulary_size)
self.extractor.setVocabulary(vocabulary)
data = []
responses = []
for id in range(len(files_array)):
data_temp, responses_temp = self.extract_train_data(files_array[id], id)
data += data_temp
responses += responses_temp
self.classifier = self.train_classifier(data, responses)
def error(self, file_list, category):
responses = np.array([self.predict(file) for file in file_list])
_responses = np.array([category for _ in range(len(responses))])
return 1 - np.sum(responses == _responses) / len(responses)
def get_images_from_folder(folder):
return ["%s/%s" % (folder, f) for f in listdir(folder) if isfile(join(folder, f))]
def start(folders, detector_type, voc_size, train_proportion):
if detector_type == "SIFT":
# "Scale Invariant Feature Transform"
extract = cv2.xfeatures2d.SIFT_create()
detector = cv2.xfeatures2d.SIFT_create()
else:
# "Speeded up Robust Features"
extract = cv2.xfeatures2d.SURF_create()
detector = cv2.xfeatures2d.SURF_create()
flann_params = dict(algorithm=1, trees=5)
matcher = cv2.FlannBasedMatcher(flann_params, {})
extractor = cv2.BOWImgDescriptorExtractor(extract, matcher)
train = []
test = []
for folder in folders:
images = get_images_from_folder(folder)
np.random.shuffle(images)
slice = int(len(images) * train_proportion)
train_images = images[0:slice]
test_images = images[slice:]
train.append(train_images)
test.append(test_images)
app = Application(extractor, detector)
app.train(train, voc_size)
total_error = 0.0
for id in range(len(test)):
print(app.error(train[id], id))
test_error = app.error(test[id], id)
print(test_error)
print("---------")
total_error = total_error + test_error
total_error = total_error / float(len(test))
print("Total error = %f" % total_error)
firstFolder = sys.argv[1]
secondFolder = sys.argv[2]
detectorType = sys.argv[3]
vocSize = int(sys.argv[4])
trainProportion = float(sys.argv[5])
start([firstFolder, secondFolder], detectorType, vocSize, trainProportion)
