diff --git a/paddle/fluid/operators/interpolate_op.cc b/paddle/fluid/operators/interpolate_op.cc
index 7ad3314e8b007bf69be1ec53ef849cf6d7593ee1..5c36048eed2b2430689fb7eb54e4cb10caa197ff 100644
--- a/paddle/fluid/operators/interpolate_op.cc
+++ b/paddle/fluid/operators/interpolate_op.cc
@@ -41,8 +41,9 @@ class InterpolateOp : public framework::OperatorWithKernel {
     int out_h, out_w;
     float scale = ctx->Attrs().Get<float>("scale");
     if (scale > 0) {
-      out_h = dim_x[2] * scale;
-      out_w = dim_x[3] * scale;
+      // round down
+      out_h = static_cast<int>(dim_x[2] * scale);
+      out_w = static_cast<int>(dim_x[3] * scale);
     } else {
       out_h = ctx->Attrs().Get<int>("out_h");
       out_w = ctx->Attrs().Get<int>("out_w");
diff --git a/paddle/fluid/operators/interpolate_op.h b/paddle/fluid/operators/interpolate_op.h
index dab626975659fed82b21754af6f18e7769ef6604..5fd42809dfec6dd821c9b27bc97d61de94b5d326 100644
--- a/paddle/fluid/operators/interpolate_op.h
+++ b/paddle/fluid/operators/interpolate_op.h
@@ -174,8 +174,8 @@ class InterpolateKernel : public framework::OpKernel<T> {
 
     float scale = ctx.Attr<float>("scale");
     if (scale > 0) {
-      out_h = in_h * scale;
-      out_w = in_w * scale;
+      out_h = static_cast<int>(in_h * scale);
+      out_w = static_cast<int>(in_w * scale);
     }
 
     auto out_size = ctx.Input<Tensor>("OutSize");
@@ -239,8 +239,8 @@ class InterpolateGradKernel : public framework::OpKernel<T> {
 
     float scale = ctx.Attr<float>("scale");
     if (scale > 0) {
-      out_h = in_h * scale;
-      out_w = in_w * scale;
+      out_h = static_cast<int>(in_h * scale);
+      out_w = static_cast<int>(in_w * scale);
     }
 
     auto out_size = ctx.Input<Tensor>("OutSize");