diff --git a/paddleslim/analysis/flops.py b/paddleslim/analysis/flops.py
index b9f19926a4ce43fcdccf3afd2e00f8d0bbf31d4d..273d7b8ff44f9eafdfe2632a383cc3dff389d8b8 100644
--- a/paddleslim/analysis/flops.py
+++ b/paddleslim/analysis/flops.py
@@ -35,13 +35,11 @@ def _graph_flops(graph, only_conv=False, detail=False):
     for op in graph.ops():
         if op.type() in ['conv2d', 'depthwise_conv2d']:
             filter_shape = op.inputs("Filter")[0].shape()
-            input_shape = op.inputs("Input")[0].shape()
             output_shape = op.outputs("Output")[0].shape()
-            _, c_in, _, _ = input_shape
-            c_out, _, k_h, k_w = filter_shape
+            c_out, c_in, k_h, k_w = filter_shape
             _, _, h_out, w_out = output_shape
-            groups = op.attr("groups")
-            kernel_ops = k_h * k_w * (float(c_in) / groups)
+            # c_in is the channel number of filter. It is (input_channel // groups).
+            kernel_ops = k_h * k_w * float(c_in)
             if len(op.inputs("Bias")) > 0:
                 with_bias = 1
             else:
@@ -50,7 +48,6 @@ def _graph_flops(graph, only_conv=False, detail=False):
             flops += op_flops
             params2flops[op.inputs("Filter")[0].name()] = op_flops
         elif op.type() == 'pool2d' and not only_conv:
-            input_shape = op.inputs("X")[0].shape()
             output_shape = op.outputs("Out")[0].shape()
             _, c_out, h_out, w_out = output_shape
             k_size = op.attr("ksize")