Definition at line 58 of file dnn_backend_native.h.

Referenced by execute_model_native(), ff_dnn_free_model_native(), ff_dnn_load_model_native(), and load_native_model().

a layer can have multiple inputs and one output.

4 is just a big enough number for input operands (increase it if necessary), do not use 'int32_t *input_operand_indexes', so we don't worry about mem leaks.

Definition at line 64 of file dnn_backend_native.h.

Referenced by execute_model_native(), ff_dnn_load_layer_avg_pool(), ff_dnn_load_layer_conv2d(), ff_dnn_load_layer_dense(), ff_dnn_load_layer_depth2space(), ff_dnn_load_layer_math_binary(), ff_dnn_load_layer_math_unary(), ff_dnn_load_layer_maximum(), and ff_dnn_load_layer_pad().

Definition at line 65 of file dnn_backend_native.h.

Referenced by execute_model_native(), ff_dnn_load_layer_avg_pool(), ff_dnn_load_layer_conv2d(), ff_dnn_load_layer_dense(), ff_dnn_load_layer_depth2space(), ff_dnn_load_layer_math_binary(), ff_dnn_load_layer_math_unary(), ff_dnn_load_layer_maximum(), and ff_dnn_load_layer_pad().

Definition at line 66 of file dnn_backend_native.h.

Referenced by execute_model_native(), ff_dnn_free_model_native(), ff_dnn_load_layer_avg_pool(), ff_dnn_load_layer_conv2d(), ff_dnn_load_layer_dense(), ff_dnn_load_layer_depth2space(), ff_dnn_load_layer_math_binary(), ff_dnn_load_layer_math_unary(), ff_dnn_load_layer_maximum(), ff_dnn_load_layer_pad(), and load_native_model().