Question

我的代码有问题，这是我的代码列表

import os
os.getcwd()
from SimpleCV import EdgeHistogramFeatureExtractor, HueHistogramFeatureExtractor, MorphologyFeatureExtractor
from SimpleCV.MachineLearning import *
from SimpleCV.ImageClass import Image, ImageSet, ColorSpace
edgeExtractor = EdgeHistogramFeatureExtractor(20) 
hueExtractor = HueHistogramFeatureExtractor(10) 
morphoExtractor = MorphologyFeatureExtractor()
extractorsList = [edgeExtractor, hueExtractor, morphoExtractor]
svm = SVMClassifier(extractorsList)
tree = TreeClassifier(extractorsList, flavor='Boosted')
naiveBayes = NaiveBayesClassifier(extractorsList)
trainPaths = ['C:\\Users\\abi\\Documents\\program coba\\ImageTrain', 'C:\\Users\\abi\\Documents\\program coba\\ImageTest']
trainPaths = ['C:\\Users\\abi\\Documents\\program coba\\ImageTrain\\Positive1', 'C:\\Users\\abi\\Documents\\program coba\\ImageTrain\\Negative1']
classes = ['Negative', 'Positive']
svm.train(trainPaths, classes, savedata="mydata.txt", verbose=False)
testPaths = ['C:\\Users\\abi\\Documents\\program coba\\ImageTest\\Positive1', 'C:\\Users\\abi\\Documents\\program coba\\ImageTest\\Negative1']
print "SVM:", svm.test(testPaths, classes, verbose=False) # [good, bad, confusion]

edgeExtractor = EdgeHistogramFeatureExtractor(20)
svm = SVMClassifier(extractorsList)

listImages = ImageSet()
   for p in ['C:\\Users\\abi\\Documents\\program coba\\NewImages\\coba']: 
        listImages += ImageSet(p)

   for image in listImages:
        className = svm.classify(image)
        image.drawText(className,25,25,fontsize=60)       
listImages.show()

之后我遇到了一些像这样的问题

Traceback (most recent call last):
  File "C:\Users\abi\Documents\program coba\imnew", line 28, in <module>
    className = svm.classify(image)
  File "C:\Python27\lib\site-packages\SimpleCV\MachineLearning\SVMClassifier.py", line 153, in classify
       featureVector.extend([self.mClassNames[0]])
IndexError: list index out of range

有IndexError按摩，但我不知道代码是什么错误。现在我使用Python 2.7.3。请帮帮我

Answer 1

svm初始化两次。第二次，您跳过svm.train()，这就是self.mClassNames[0]为空的原因。

mClassNames由classes参数填充：

svm.train(trainPaths, classes, savedata="mydata.txt", verbose=False)`

只需删除第二个svm = SVMClassifier(extractorsList)即可解析IndexError。

但请注意，根据documentation，您的classes订单与trainPaths中的图片路径顺序不一致：

图像 - 路径或图像集的顺序必须与类类型的顺序相同 - 请注意，所有图像类必须位于单独的目录中 - 类名也必须与目录对齐。

这表明classes应更改为关注trainPaths：

trainPaths = ['C:\\Users\\abi\\Documents\\program coba\\ImageTrain\\Positive1', 
              'C:\\Users\\abi\\Documents\\program coba\\ImageTrain\\Negative1']
classes = ['Positive', 'Negative']

您可能还需要让classes字符串与图像目录字符串完全匹配。源代码中的注释有点含糊不清。您还出于某种原因定义了trainPaths两次。

使用python的SVM分类器中的“IndexError：列表索引超出范围”

1 个答案: