diff --git a/openbr/plugins/cuda/copyfrom.cpp b/openbr/plugins/cuda/copyfrom.cpp
index 682e9da..aada3c2 100644
--- a/openbr/plugins/cuda/copyfrom.cpp
+++ b/openbr/plugins/cuda/copyfrom.cpp
@@ -10,8 +10,7 @@ using namespace cv;
 
 // extern CUDA declaration
 namespace br { namespace cuda { namespace cudacopyfrom {
-  //template <typename T> void wrapper(void* src, T* out, int rows, int cols) {
-  void wrapper(void* src, float* out, const int rows, const int cols);
+  template <typename T> void wrapper(void* src, T* out, int rows, int cols);
 }}}
 
 namespace br
@@ -32,26 +31,21 @@ private:
       int cols = *((int*)dataPtr[2]);
       int type = *((int*)dataPtr[3]);
 
-      if (type != CV_32FC1) {
-        cout << "ERR: Invalid data type!" << endl;
-        return;
-      }
-
-      cout << "cudaMemPtr: " << cudaMemPtr << endl;
-      cout << "rows: " << rows << endl;
-      cout << "cols: " << cols << endl;
-      cout << "type: " << type << endl;
-
       Mat dstMat = Mat(rows, cols, type);
-      br::cuda::cudacopyfrom::wrapper(cudaMemPtr, dstMat.ptr<float>(), rows, cols);
+      switch(type) {
+      case CV_32FC1:
+        br::cuda::cudacopyfrom::wrapper(cudaMemPtr, dstMat.ptr<float>(), rows, cols);
+        break;
+      case CV_8UC1:
+        br::cuda::cudacopyfrom::wrapper(cudaMemPtr, dstMat.ptr<unsigned char>(), rows, cols);
+        break;
+      default:
+        cout << "ERR: Invalid image format" << endl;
+        break;
+      }
       dst = dstMat;
 
       cout << "CUDACopyFrom End" << endl;
-
-      cout << "DST Data" << endl;
-      cout << "rows: " << dstMat.rows << endl;
-      cout << "cols: " << dstMat.cols << endl;
-      cout << "type: " << dstMat.type() << endl;
     }
   };
 
diff --git a/openbr/plugins/cuda/copyfrom.cu b/openbr/plugins/cuda/copyfrom.cu
index 2b93f4b..2a85f93 100644
--- a/openbr/plugins/cuda/copyfrom.cu
+++ b/openbr/plugins/cuda/copyfrom.cu
@@ -1,7 +1,9 @@
 namespace br { namespace cuda { namespace cudacopyfrom {
-  //template <typename T> void wrapper(void* src, T* out, int rows, int cols) {
-  void wrapper(void* src, float* dst, const int rows, const int cols) {
-    cudaMemcpy(dst, src, rows*cols*sizeof(float), cudaMemcpyDeviceToHost);
+  template <typename T> void wrapper(void* src, T* dst, int rows, int cols) {
+    cudaMemcpy(dst, src, rows*cols*sizeof(T), cudaMemcpyDeviceToHost);
     cudaFree(src);
   }
+
+  template void wrapper(void*, float*, int, int);
+  template void wrapper(void*, unsigned char*, int, int);
 }}}
diff --git a/openbr/plugins/cuda/copyto.cpp b/openbr/plugins/cuda/copyto.cpp
index 3e8cd48..7288553 100644
--- a/openbr/plugins/cuda/copyto.cpp
+++ b/openbr/plugins/cuda/copyto.cpp
@@ -11,9 +11,7 @@ using namespace cv;
 extern string type2str(int type);
 
 namespace br { namespace cuda { namespace cudacopyto {
-  //template<typename T>
-  //void wrapper(const T* in, void** out, const int rows, const int cols);
-  void wrapper(const unsigned char* in, void** out, const int rows, const int cols);
+  template <typename T> void wrapper(const T* in, void** out, const int rows, const int cols);
 }}}
 
 namespace br
@@ -31,11 +29,10 @@ private:
 
       void* cudaMemPtr;
       switch(srcMat.type()) {
-      //case CV_32FC1:
-      //  br::cuda::cudacopyfrom::wrapper<float>(srcMat.ptr<float>(), &cudaMemPtr, rows, cols);
-      //  break;
+      case CV_32FC1:
+        br::cuda::cudacopyto::wrapper(srcMat.ptr<float>(), &cudaMemPtr, rows, cols);
+        break;
       case CV_8UC1:
-        //br::cuda::cudacopyfrom::wrapper<unsigned char>(srcMat.ptr<unsigned char>(), &cudaMemPtr, rows, cols);
         br::cuda::cudacopyto::wrapper(srcMat.ptr<unsigned char>(), &cudaMemPtr, rows, cols);
         break;
       default:
diff --git a/openbr/plugins/cuda/copyto.cu b/openbr/plugins/cuda/copyto.cu
index c4885a4..feecb83 100644
--- a/openbr/plugins/cuda/copyto.cu
+++ b/openbr/plugins/cuda/copyto.cu
@@ -1,8 +1,9 @@
 namespace br { namespace cuda { namespace cudacopyto {
-  //template<typename T>
-  //void wrapper(const T* in, void** out, const int rows, const int cols) {
-  void wrapper(const unsigned char* in, void** out, const int rows, const int cols) {
-    cudaMalloc(out, rows*cols*sizeof(unsigned char));
-    cudaMemcpy(*out, in, rows*cols*sizeof(unsigned char), cudaMemcpyHostToDevice);
+  template <typename T> void wrapper(const T* in, void** out, const int rows, const int cols) {
+    cudaMalloc(out, rows*cols*sizeof(T));
+    cudaMemcpy(*out, in, rows*cols*sizeof(T), cudaMemcpyHostToDevice);
   }
+
+  template void wrapper(const float* in, void** out, const int rows, const int cols);
+  template void wrapper(const unsigned char* in, void** out, const int rows, const int cols);
 }}}