public ImageTrainDataSet SplitTestData(int num)
{
Random rand = new Random();
//List<ImageTrainData> ret = _self.GetRange(0, num);
//_self.RemoveRange(0, num);
int numOutput = this.GetNumOutput();
ImageTrainDataSet ret = new ImageTrainDataSet();
while (ret.Count() < num)
{
for (int i = 0; i < numOutput; i++)
{
while (true)
{
int index = rand.Next(this.Count);
if (this[index].labelIndex == i)
{
ret.Add(this[index]);
this.RemoveAt(index);
break;
}
}
}
}
return(ret);
}
private void loadTrainDataToolStripMenuItem_Click_1(object sender, EventArgs e)
{
trainData = FeatureDetector.GetAllTrainImageData(dataPath, configure.trainFolders);
trainData = trainData.BalanceData();
trainData.Shuffle();
testData = trainData.SplitTestData((int)(0.1 * trainData.Count()));
TrainDataInfo[] infos = trainData.GetInfo(configure.trainFolders.Count());
TrainDataInfo[] testinfos = testData.GetInfo(configure.trainFolders.Count());
logger.logStr(Utils.ToJsonString(infos, true));
logger.logStr(Utils.ToJsonString(testinfos, true));
}
public int Evaluate(ImageTrainDataSet trainData)
{
double[][] features = trainData.GetFeature(bow, mask);
int[] labelIndexs = trainData.GetLabelIndexs();
String[] labels = trainData.GetLabels();
int errorCount = 0;
for (int i = 0; i < trainData.Count(); i++)
{
//double[] feature = bow.Transform(images[i]);
double[] answer = network.Compute(features[i]);
int expected = labelIndexs[i];
int actual; answer.Max(out actual);
if (actual != expected)
{
errorCount++;
}
}
return(errorCount);
}