diff --git a/paddle/operators/iou_similarity_op.cc b/paddle/operators/iou_similarity_op.cc
index 49e54a1a53c4df7450ec6159d7d58938593a2f06..6c1b99b46f288b9dea94727b23a2342137a33d27 100755
--- a/paddle/operators/iou_similarity_op.cc
+++ b/paddle/operators/iou_similarity_op.cc
@@ -44,11 +44,14 @@ class IOUSimilarityOpMaker : public framework::OpProtoAndCheckerMaker {
   IOUSimilarityOpMaker(OpProto *proto, OpAttrChecker *op_checker)
       : OpProtoAndCheckerMaker(proto, op_checker) {
     AddInput("X",
-             "(Tensor, default Tensor<float>) "
-             "Box list X holds N boxes, each box is "
-             "represented as [xmin, ymin, xmax, ymax], the shape of X is [N, "
-             "4]. [xmin, ymin] is the lower left coordinate of the box, and "
-             "[xmax, ymax] is the right upper coordinate of the box.");
+             "(LoDTensor, default LoDTensor<float>) "
+             "Box list X is a 2-D LoDTensor with shape [N, 4] holds N boxes, "
+             "each box is represented as [xmin, ymin, xmax, ymax], "
+             "the shape of X is [N, 4]. [xmin, ymin] is the lower left "
+             "coordinate of the box, and [xmax, ymax] is the right upper "
+             "coordinate of the box.This tensor can contain LoD information "
+             "to represent a batch of inputs. One instance of this batch can "
+             "contain different numbers of entities.");
     AddInput("Y",
              "(Tensor, default Tensor<float>) "
              "Box list Y holds M boxes, each box is "
@@ -56,14 +59,23 @@ class IOUSimilarityOpMaker : public framework::OpProtoAndCheckerMaker {
              "4]. [xmin, ymin] is the lower left coordinate of the box, and "
              "[xmax, ymax] is the right upper coordinate of the box.");
 
-    AddOutput(
-        "Out",
-        "(Tensor) The output of iou_similarity op, a tensor with shape [N, M] "
-        "representing pairwise iou scores.");
+    AddOutput("Out",
+              "(LoDTensor or Tensor, the lod is same as input X) The output of "
+              "iou_similarity op, a tensor with shape [N, M] "
+              "representing pairwise iou scores.");
 
     AddComment(R"DOC(
 IOU Similarity Operator.
 Computes intersection-over-union (IOU) between two box lists.
+ Box list 'X' should be a LoDTensor and 'Y' is a common Tensor,
+ boxes in 'Y' are shared by all input images.
+ Given two box A and B, the calculation of IOU is as follows:
+
+$$
+IOU(A, B) = 
+\frac{area(A\cap B)}{area(A)+area(B)-area(A\cap B)}
+$$
+
 )DOC");
   }
 };
diff --git a/paddle/operators/iou_similarity_op.cu b/paddle/operators/iou_similarity_op.cu
index 526edaefeea98b0760832c6b24fcb2d38b3aa00e..fa5052624618c35875b241419946f69b776c81d4 100755
--- a/paddle/operators/iou_similarity_op.cu
+++ b/paddle/operators/iou_similarity_op.cu
@@ -12,7 +12,6 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 See the License for the specific language governing permissions and
 limitations under the License. */
 
-#define EIGEN_USE_GPU
 #include "paddle/operators/iou_similarity_op.h"
 
 namespace ops = paddle::operators;
diff --git a/paddle/operators/iou_similarity_op.h b/paddle/operators/iou_similarity_op.h
index fa1b60ae34eeee2cf534607606568f527f1bd67b..f42e05b078b13f9b2f58b3bd544ba4bd881940da 100644
--- a/paddle/operators/iou_similarity_op.h
+++ b/paddle/operators/iou_similarity_op.h
@@ -71,9 +71,9 @@ template <typename DeviceContext, typename T>
 class IOUSimilarityKernel : public framework::OpKernel<T> {
  public:
   void Compute(const framework::ExecutionContext& ctx) const override {
-    const framework::Tensor* in_x = ctx.Input<framework::Tensor>("X");
+    const framework::LoDTensor* in_x = ctx.Input<framework::LoDTensor>("X");
     const framework::Tensor* in_y = ctx.Input<framework::Tensor>("Y");
-    framework::Tensor* out = ctx.Output<framework::Tensor>("Out");
+    framework::LoDTensor* out = ctx.Output<framework::LoDTensor>("Out");
 
     int x_n = in_x->dims()[0];
     int y_n = in_y->dims()[0];
@@ -83,6 +83,8 @@ class IOUSimilarityKernel : public framework::OpKernel<T> {
     platform::ForRange<DeviceContext> for_range(
         static_cast<const DeviceContext&>(ctx.device_context()), x_n);
     for_range(functor);
+
+    out->set_lod(in_x->lod());
   }
 };  // namespace operators
 
diff --git a/python/paddle/v2/fluid/tests/test_iou_similarity_op.py b/python/paddle/v2/fluid/tests/test_iou_similarity_op.py
index 694da618628f6f8986a50d62e8c4b144f298d8bd..128f2e4977195a563efcd26364cc6261da2dd685 100755
--- a/python/paddle/v2/fluid/tests/test_iou_similarity_op.py
+++ b/python/paddle/v2/fluid/tests/test_iou_similarity_op.py
@@ -38,5 +38,18 @@ class TestIOUSimilarityOp(OpTest):
         self.outputs = {'Out': self.output}
 
 
+class TestIOUSimilarityOpWithLoD(TestIOUSimilarityOp):
+    def test_check_output(self):
+        self.check_output()
+
+    def setUp(self):
+        super(TestIOUSimilarityOpWithLoD, self).setUp()
+        self.boxes1_lod = [[0, 1, 2]]
+        self.output_lod = [[0, 1, 2]]
+
+        self.inputs = {'X': (self.boxes1, self.boxes1_lod), 'Y': self.boxes2}
+        self.outputs = {'Out': (self.output, self.output_lod)}
+
+
 if __name__ == '__main__':
     unittest.main()