currently, only float is supported as model input, actually, there
are other data types, this patch adds uint8.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>