Bug #2214: Fixed bug, when calling the train method multiple times. The Eigenfaces and Fisherfaces method now re-estimate the model, the LBPH method appends the new data to the existing model.

modules/contrib/src/facerec.cpp

@@ -338,6 +338,9 @@ void Eigenfaces::train(InputArray _src, InputArray _local_labels) {
         string error_message = format("The number of samples (src) must equal the number of labels (labels)! len(src)=%d, len(labels)=%d.", n, labels.total());
         CV_Error(CV_StsBadArg, error_message);
     }
+    // clear existing model data
+    _labels.release();
+    _projections.clear();
     // clip number of components to be valid
     if((_num_components <= 0) || (_num_components > n))
         _num_components = n;
@@ -347,7 +350,8 @@ void Eigenfaces::train(InputArray _src, InputArray _local_labels) {
     _mean = pca.mean.reshape(1,1); // store the mean vector
     _eigenvalues = pca.eigenvalues.clone(); // eigenvalues by row
     transpose(pca.eigenvectors, _eigenvectors); // eigenvectors by column
-    labels.copyTo(_labels); // store labels for prediction
+    // store labels for prediction
+    labels.copyTo(_labels);
     // save projections
     for(int sampleIdx = 0; sampleIdx < data.rows; sampleIdx++) {
         Mat p = subspaceProject(_eigenvectors, _mean, data.row(sampleIdx));
@@ -443,7 +447,10 @@ void Fisherfaces::train(InputArray src, InputArray _lbls) {
         string error_message = format("Expected the labels in a matrix with one row or column! Given dimensions are rows=%s, cols=%d.", labels.rows, labels.cols);
        CV_Error(CV_StsBadArg, error_message);
     }
-    // Get the number of unique classes (provide a cv::Mat overloaded version?)
+    // clear existing model data
+    _labels.release();
+    _projections.clear();
+    // get the number of unique classes (provide a cv::Mat overloaded version?)
     vector<int> ll;
     labels.copyTo(ll);
     int C = (int) remove_dups(ll).size();
@@ -462,7 +469,7 @@ void Fisherfaces::train(InputArray src, InputArray _lbls) {
     lda.eigenvalues().convertTo(_eigenvalues, CV_64FC1);
     // Now calculate the projection matrix as pca.eigenvectors * lda.eigenvectors.
     // Note: OpenCV stores the eigenvectors by row, so we need to transpose it!
-    gemm(pca.eigenvectors, lda.eigenvectors(), 1.0, Mat(), 0.0, _eigenvectors, CV_GEMM_A_T);
+    gemm(pca.eigenvectors, lda.eigenvectors(), 1.0, Mat(), 0.0, _eigenvectors, GEMM_1_T);
     // store the projections of the original data
     for(int sampleIdx = 0; sampleIdx < data.rows; sampleIdx++) {
         Mat p = subspaceProject(_eigenvectors, _mean, data.row(sampleIdx));
@@ -525,6 +532,7 @@ void Fisherfaces::save(FileStorage& fs) const {
     writeFileNodeList(fs, "projections", _projections);
     fs << "labels" << _labels;
 }
+
 //------------------------------------------------------------------------------
 // LBPH
 //------------------------------------------------------------------------------
@@ -724,7 +732,8 @@ void LBPH::train(InputArray _src, InputArray _lbls) {
     if(_src.kind() != _InputArray::STD_VECTOR_MAT && _src.kind() != _InputArray::STD_VECTOR_VECTOR) {
         string error_message = "The images are expected as InputArray::STD_VECTOR_MAT (a std::vector<Mat>) or _InputArray::STD_VECTOR_VECTOR (a std::vector< vector<...> >).";
         CV_Error(CV_StsBadArg, error_message);
-    } else  if(_src.total() == 0) {
+    }
+    if(_src.total() == 0) {
         string error_message = format("Empty training data was given. You'll need more than one sample to learn a model.");
         CV_Error(CV_StsUnsupportedFormat, error_message);
     } else if(_lbls.getMat().type() != CV_32SC1) {
@@ -734,22 +743,19 @@ void LBPH::train(InputArray _src, InputArray _lbls) {
     // get the vector of matrices
     vector<Mat> src;
     _src.getMatVector(src);
-    for (vector<Mat>::const_iterator image = src.begin(); image != src.end(); ++image) {
+    // get the label matrix
-      if (image->channels() != 1) {
-        string error_message = format("The images must be single channel (grayscale), but an image has %d channels.", image->channels());
-	CV_Error(CV_StsUnsupportedFormat, error_message);
-      }
-    }
-    // turn the label matrix into a vector
     Mat labels = _lbls.getMat();
-    CV_Assert( labels.type() == CV_32S && (labels.cols == 1 || labels.rows == 1));
+    // check if data is well- aligned
     if(labels.total() != src.size()) {
-        CV_Error(CV_StsUnsupportedFormat, "The number of labels must equal the number of samples.");
+        string error_message = format("The number of samples (src) must equal the number of labels (labels). Was len(samples)=%d, len(labels)=%d.", src.size(), _labels.total());
+        CV_Error(CV_StsBadArg, error_message);
+    }
+    // append labels to _labels matrix
+    for(int labelIdx = 0; labelIdx < labels.total(); labelIdx++) {
+        _labels.push_back(labels.at<int>(labelIdx));
     }
-    // store given labels
-    labels.copyTo(_labels);
     // store the spatial histograms of the original data
-    for(size_t sampleIdx = 0; sampleIdx < src.size(); sampleIdx++) {
+    for(int sampleIdx = 0; sampleIdx < src.size(); sampleIdx++) {
         // calculate lbp image
         Mat lbp_image = elbp(src[sampleIdx], _radius, _neighbors);
         // get spatial histogram from this lbp image
@@ -765,11 +771,12 @@ void LBPH::train(InputArray _src, InputArray _lbls) {
 }
 
 void LBPH::predict(InputArray _src, int &minClass, double &minDist) const {
-    Mat src = _src.getMat();
+    if(_histograms.empty()) {
-    if (src.channels() != 1) {
+        // throw error if no data (or simply return -1?)
-      string error_message = format("The image must be single channel (grayscale), but the image has %d channels.", src.channels());
+        string error_message = "This LBPH model is not computed yet. Did you call the train method?";
-      CV_Error(CV_StsUnsupportedFormat, error_message);
+        CV_Error(CV_StsBadArg, error_message);
     }
+    Mat src = _src.getMat();
     // get the spatial histogram from input image
     Mat lbp_image = elbp(src, _radius, _neighbors);
     Mat query = spatial_histogram(
@@ -781,11 +788,11 @@ void LBPH::predict(InputArray _src, int &minClass, double &minDist) const {
     // find 1-nearest neighbor
     minDist = DBL_MAX;
     minClass = -1;
-    for(size_t sampleIdx = 0; sampleIdx < _histograms.size(); sampleIdx++) {
+    for(int sampleIdx = 0; sampleIdx < _histograms.size(); sampleIdx++) {
         double dist = compareHist(_histograms[sampleIdx], query, CV_COMP_CHISQR);
         if((dist < minDist) && (dist < _threshold)) {
             minDist = dist;
-            minClass = _labels.at<int>((int)sampleIdx);
+            minClass = _labels.at<int>(sampleIdx);
         }
     }
 }