利用jetson orin 部署后程序执行无输出

[jalywang123]

OpenCV4.5以及第三方插件编译成GPU支持的，MNN为2.5版本，这两个均编译顺利，但执行./PlateRecDemo ../hyperlpr3/resource/models/r2_mobile ../hyperlpr3/resource/images/test_img.jpg后输出如下： [InferenceHelper][133] Use MNN The device support i8sdot:1, support fp16:1, support i8mm: 0 [InferenceHelper][133] Use MNN [InferenceHelper][133] Use MNN [InferenceHelper][133] Use MNN 后执行单独模块测试，出现报错如下： /host/tw_back/HyperLPR/cpp/test/nn_module/test_classification.cpp:48: FAILED: CHECK( PlateColor(clsEngine.getMOutputColor()) == predict_results_cls[i] ) with expansion: 1 == 0

/host/tw_back/HyperLPR/cpp/test/nn_module/test_classification.cpp:49: FAILED: CHECK( clsEngine.getMOutputMaxConfidence() == Approx(predict_results_confidence[i]).epsilon(0.001) ) with expansion: 0.06262f == Approx( 0.9999293089 )

/host/tw_back/HyperLPR/cpp/test/nn_module/test_classification.cpp:48: FAILED: CHECK( PlateColor(clsEngine.getMOutputColor()) == predict_results_cls[i] ) with expansion: 1 == 2

/host/tw_back/HyperLPR/cpp/test/nn_module/test_classification.cpp:49: FAILED: CHECK( clsEngine.getMOutputMaxConfidence() == Approx(predict_results_confidence[i]).epsilon(0.001) ) with expansion: 0.06262f == Approx( 0.8975974917 )

/host/tw_back/HyperLPR/cpp/test/nn_module/test_classification.cpp:49: FAILED: CHECK( clsEngine.getMOutputMaxConfidence() == Approx(predict_results_confidence[i]).epsilon(0.001) ) with expansion: 0.06262f == Approx( 0.9997951984 )

=============================================================================== [test_detection.cpp][C_A_T_C_H_T_E_S_T_0][15]: [UnitTest]->Detect Model [test_detection.cpp][C_A_T_C_H_T_E_S_T_0][25]: Detect Model SplitModel [InferenceHelper][133] Use MNN [InferenceHelper][133] Use MNN

test_Detection test_SplitDetectionSplitModel

/host/tw_back/HyperLPR/cpp/test/nn_module/test_detection.cpp:24 ...............................................................................

/host/tw_back/HyperLPR/cpp/test/nn_module/test_detection.cpp:32: FAILED: CHECK( result.size() == 1 ) with expansion: 0 == 1

/host/tw_back/HyperLPR/cpp/test/nn_module/test_detection.cpp:32: FAILED: {Unknown expression after the reported line} due to a fatal error condition: SIGSEGV - Segmentation violation signal

=============================================================================== test cases: 2 | 0 passed | 2 failed assertions: 26 | 19 passed | 7 failed 麻烦给看一下可能出问题的地方

[tunmx]

看日志是推理结果与预期结果的预期不一致导致的，可能与三方库版本关系较大，我手头上没有jetson设备，稍后我使用MNN2.5版本在本地尝试一下

[lzylzylzy123456]

orin nx是有输出的，是不是MNN版本太低了。