From: Inki Dae <inki.dae@samsung.com>
Date: Mon, 7 Jun 2021 07:44:29 +0000 (+0900)
Subject: Fix backend type issue
X-Git-Tag: submit/tizen/20210817.030804~5
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=e7d8e9adae10560f2410068a19c75e6094642657;p=platform%2Fcore%2Fmultimedia%2Finference-engine-interface.git

Fix backend type issue

Fixed a backend issue that when user requested an inference with
ONE type, it passed TFLITE type instead of ONE.

So this patch passes ONE type to inference-engine-mlapi backend correctly.

Change-Id: Ib91e9880873f1023dfb7bc8bf3b5f838029527a3
Signed-off-by: Inki Dae <inki.dae@samsung.com>
---

diff --git a/src/inference_engine_common_impl.cpp b/src/inference_engine_common_impl.cpp
index f28e67b..5f709a2 100644
--- a/src/inference_engine_common_impl.cpp
+++ b/src/inference_engine_common_impl.cpp
@@ -328,10 +328,10 @@ namespace Common
 		}
 
 		// If user requested an inference with ONE backend and CPU/GPU device then
-		// pass TFLITE backend type to MLAPI backend.
+		// pass ONE backend type to MLAPI backend.
 		if (backend_type == INFERENCE_BACKEND_ONE &&
 				(device_type & INFERENCE_TARGET_CPU || device_type & INFERENCE_TARGET_GPU))
-			backend_type = INFERENCE_BACKEND_TFLITE;
+			backend_type = INFERENCE_BACKEND_ONE;
 
 		// If NPU type is declared in ini file then pass the type to
 		// a given inference engine backend.