[quantizer] reduce the number of pad operators (#381)

* [quantizer] reduce the number of pad operators

* rename add_padop to add_pad_op

tinynn/converter/operators/torch/aten.py

@@ -1386,11 +1386,17 @@ def parse(self, node, attrs, args, graph_converter):
 
         assert dilation_h == dilation_w == 1, "Only dilation == 1 is supported"
 
-        padding = tfl_schema.Padding.VALID
+        add_pad_op = not (
+            stride_h == stride_w == 1 and pad_h == kernel_h // 2 and pad_w == kernel_w // 2 and not ceil_mode
+        )
+        padding = tfl_schema.Padding.SAME
+        if add_pad_op:
+            padding = tfl_schema.Padding.VALID
 
         maxpool_op = tfl.MaxPool2dOperator(inputs, outputs, padding, stride_w, stride_h, kernel_w, kernel_h)
         ops = self.wrap_ops_with_nhwc_nchw_transposes([maxpool_op])
-        self.handle_padding(pad_h, pad_w, 1, ops, ceil_mode)
+        if add_pad_op:
+            self.handle_padding(pad_h, pad_w, 1, ops, ceil_mode)
 
         for op in ops:
             graph_converter.add_operator(op)