diff --git a/openbr/plugins/cuda/cudacvt.cpp b/openbr/plugins/cuda/cudacvt.cpp
index 28e4999..1f3297d 100644
--- a/openbr/plugins/cuda/cudacvt.cpp
+++ b/openbr/plugins/cuda/cudacvt.cpp
@@ -70,7 +70,7 @@ private:
         std::cout << "rows: " << rows << std::endl;
         std::cout << "cols: " << cols << std::endl;
 
-        Mat dstMat = Mat(src.m().rows, src.m().cols, CV_8UC1);
+        Mat dstMat = Mat(src.m().rows, src.m().cols, src.m().type());
         void** dstDataPtr = dstMat.ptr<void*>();
         dstDataPtr[1] = srcDataPtr[1];
         dstDataPtr[2] = srcDataPtr[2];
diff --git a/openbr/plugins/cuda/cudacvt.cu b/openbr/plugins/cuda/cudacvt.cu
index 372c36f..9c07c6b 100644
--- a/openbr/plugins/cuda/cudacvt.cu
+++ b/openbr/plugins/cuda/cudacvt.cu
@@ -27,7 +27,7 @@ namespace br{ namespace cuda {
       }
     }
 
-    dstPtr[index] = 0;
+    dstPtr[index] = 0.299f * srcPtr[3*index] + 0.587f * srcPtr[3*index+1] + 0.114f * srcPtr[3*index+2];
     return;
   }
 
@@ -41,7 +41,7 @@ namespace br{ namespace cuda {
     CUDA_SAFE_MALLOC(dstPtr, rows*cols*sizeof(uint8_t), &err);
     std::cout << "After malloc" << std::endl;
 
-    //cudacvt_kernel<<<numBlocks, threadsPerBlock>>>((uint8_t*)srcPtr, (uint8_t*) (*dstPtr), rows, cols);
+    cudacvt_kernel<<<numBlocks, threadsPerBlock>>>((uint8_t*)srcPtr, (uint8_t*) (*dstPtr), rows, cols);
     CUDA_KERNEL_ERR_CHK(&err);
     CUDA_SAFE_FREE(srcPtr, &err);
   }