diff --git a/modules/gapi/include/opencv2/gapi/media.hpp b/modules/gapi/include/opencv2/gapi/media.hpp
index f27cb80913..3d7f5a5b65 100644
--- a/modules/gapi/include/opencv2/gapi/media.hpp
+++ b/modules/gapi/include/opencv2/gapi/media.hpp
@@ -30,9 +30,21 @@ public:
     View access(Access) const;
     cv::GFrameDesc desc() const;
 
+    // Cast underlying MediaFrame adapter to the particular adapter type,
+    // return nullptr if underlying type is different
+    template<typename T> T* get() const
+    {
+        static_assert(std::is_base_of<IAdapter, T>::value,
+                      "T is not derived from cv::MediaFrame::IAdapter!");
+        auto* adapter = getAdapter();
+        GAPI_Assert(adapter != nullptr);
+        return dynamic_cast<T*>(adapter);
+    }
+
 private:
     struct Priv;
     std::shared_ptr<Priv> m;
+    IAdapter* getAdapter() const;
 };
 
 template<class T, class... Args>
diff --git a/modules/gapi/src/api/media.cpp b/modules/gapi/src/api/media.cpp
index 212902ee3b..3c171b215f 100644
--- a/modules/gapi/src/api/media.cpp
+++ b/modules/gapi/src/api/media.cpp
@@ -26,6 +26,10 @@ cv::MediaFrame::View cv::MediaFrame::access(Access code) const {
     return m->adapter->access(code);
 }
 
+cv::MediaFrame::IAdapter* cv::MediaFrame::getAdapter() const {
+    return m->adapter.get();
+}
+
 cv::MediaFrame::View::View(Ptrs&& ptrs, Strides&& strs, Callback &&cb)
     : ptr   (std::move(ptrs))
     , stride(std::move(strs))
diff --git a/modules/gapi/src/backends/ie/giebackend.cpp b/modules/gapi/src/backends/ie/giebackend.cpp
index 2a26f0c329..3af09d4381 100644
--- a/modules/gapi/src/backends/ie/giebackend.cpp
+++ b/modules/gapi/src/backends/ie/giebackend.cpp
@@ -126,25 +126,20 @@ inline IE::TensorDesc toIE(const cv::Mat &mat, cv::gapi::ie::TraitAs hint) {
     if (sz.dims() == 2 && hint == cv::gapi::ie::TraitAs::IMAGE)
     {
         // NB: This logic is mainly taken from IE samples
-        const size_t pixsz    = CV_ELEM_SIZE1(mat.type());
         const size_t channels = mat.channels();
         const size_t height   = mat.size().height;
         const size_t width    = mat.size().width;
 
-        const size_t strideH  = mat.step.buf[0];
-        const size_t strideW  = mat.step.buf[1];
+        const size_t strideH  = mat.step[0];
 
-        const bool is_dense =
-            strideW == pixsz * channels &&
-            strideH == strideW * width;
-
-        if (!is_dense)
-            cv::util::throw_error(std::logic_error("Doesn't support conversion"
-                                                   " from non-dense cv::Mat"));
+        IE::BlockingDesc bdesc({1, height, width, channels} /* dims */,
+                               {0, 2, 3, 1} /* order for NHWC */,
+                               0 /* offset */,
+                               {0, 0, 0, 0} /* offsets for dims */,
+                               {strideH * height, strideH, channels, 1} /* strides for dims */);
 
         return IE::TensorDesc(toIE(mat.depth()),
-                              IE::SizeVector{1, channels, height, width},
-                              IE::Layout::NHWC);
+                              IE::SizeVector{1, channels, height, width}, bdesc);
     }
 
     return IE::TensorDesc(toIE(mat.depth()), toIE(sz), toIELayout(sz.dims()));