I have to do a project that recognize images using opencv machine learning. I should try many methods. I tried KNN and SVM but i don't know how to implement ANN. Can someone help me?
This is the KNN example
public void train() {
// Creating Training Data
Mat trainData = new Mat();
Mat train_labels = new Mat();
for (int i = 0; i <6; i++) {
String path = Environment.getExternalStorageDirectory().toString()
+ "/Pictures/images/" + i + ".jpg";
Mat img = Imgcodecs.imread(path);
Log.i(TAG,"error"+i+img.empty());
img.convertTo(img, CvType.CV_32FC1); // Convert to float
Size dsize = new Size(25, 25);
Imgproc.resize(img, img, dsize);
img.convertTo(img, CvType.CV_32FC1);
Mat imgResized = img.reshape(1, 1); // make continuous
trainData.push_back(imgResized);
// add 1 item
train_labels
.push_back(new Mat(1, 1, CvType.CV_32SC1, new Scalar(i)));
}
Mat response = new Mat();
Mat tmp;
tmp = train_labels.reshape(1, 1); // make continuous
tmp.convertTo(response, CvType.CV_32FC1); // Convert to float
KNearest knn = KNearest.create();
knn.train(trainData, Ml.ROW_SAMPLE, train_labels);
// For Storing training data
String path = Environment.getExternalStoragePublicDirectory(
Environment.DIRECTORY_DOWNLOADS).toString();
Bitmap bmpOut1 = Bitmap.createBitmap(trainData.cols(),
trainData.rows(), Bitmap.Config.ARGB_8888);
Bitmap bmpOut2 = Bitmap.createBitmap(response.cols(), response.rows(),
Bitmap.Config.ARGB_8888);
response.convertTo(response, CvType.CV_8UC1);
trainData.convertTo(trainData, CvType.CV_8UC1);
Utils.matToBitmap(trainData, bmpOut1);
Utils.matToBitmap(response, bmpOut2);
// File file = new File(path);
// file.mkdirs();
File file = new File(path, "train.png");
File file2 = new File(path, "response.png");
OutputStream fout = null;
OutputStream fout2 = null;
try {
fout = new FileOutputStream(file);
fout2 = new FileOutputStream(file2);
BufferedOutputStream bos = new BufferedOutputStream(fout);
BufferedOutputStream bos2 = new BufferedOutputStream(fout2);
bmpOut1.compress(Bitmap.CompressFormat.PNG, 100, bos);
bos.flush();
bos.close();
bmpOut1.recycle();
bmpOut2.compress(Bitmap.CompressFormat.PNG, 100, bos2);
bos2.flush();
bos2.close();
bmpOut2.recycle();
} catch (FileNotFoundException e) {
e.printStackTrace();
}
catch (IOException e) {
e.printStackTrace();
}
// For Accessing training data in BMP file
BitmapFactory.Options o = new BitmapFactory.Options();
o.inScaled = false;
Bitmap blankBitmap = BitmapFactory.decodeResource(getResources(),
R.drawable.train, o);
Mat trainData2 = new Mat();
Utils.bitmapToMat(blankBitmap, trainData2);
Bitmap blankBitmap2 = BitmapFactory.decodeResource(getResources(),
R.drawable.response, o);
Mat response2 = new Mat();
Utils.bitmapToMat(blankBitmap2, response2);
Mat response3 = new Mat();
Mat trainData3 = new Mat();
// 1. change the number of channels
Imgproc.cvtColor(response2, response2, Imgproc.COLOR_BGRA2GRAY);
Imgproc.cvtColor(trainData2, trainData2, Imgproc.COLOR_BGRA2GRAY);
response2.convertTo(response3, CvType.CV_32FC1);
trainData2.convertTo(trainData3, CvType.CV_32FC1);
// instantiate the KNN object
KNearest knn2 = KNearest.create();
knn2.train(trainData3, Ml.ROW_SAMPLE, response3);
Mat img = Imgcodecs.imread("/storage/emulated/0/DCIM/Camera/IMG.jpg");
Log.i(TAG, "error1 img" + img.empty());
Mat test = new Mat();
Imgproc.resize(img, test, new Size(25, 25));
test.convertTo(test, CvType.CV_32FC1);
Mat results = new Mat();
Mat responses = new Mat();
Mat dists = new Mat();
float label = knn.findNearest(test.reshape(1, 1),1, results,responses, dists);
if (label==0)
{
text.setText("Le reve");
}
else if (label==1) {
text.setText("Guernica");
}
else if (label==2) {
text.setText("La cène");
}
else if (label==3) {
text.setText("La joconde");
}
else if (label==4) {
text.setText("la vierge et sainte anne");
}
else if (label==5) {
text.setText("les joueurs de cartes ");
}
else
text.setText("");
}
↧