detector = trainFaste​rRCNNObjec​tDetector(​trainData, layers, options, ... 'NegativeOverlapRange', [0 0.3], ... 'PositiveOverlapRange', [0.6 1], ... 'BoxPyramidScale', 1.2);

Error: parallel.gpu.CUDADevice.hBuild

An unexpected error occurred trying to retrieve CUDA device properties. The CUDA error was:

CUDA_ERROR_UNKNOWN

Error: parallel.gpu.GPUDevice.getDevice (line 76)

d = parallel.gpu.CUDADevice.hBuild( idx );

Error: parallel.gpu.GPUDevice.isAvailable (line 146)

device = parallel.gpu.GPUDevice.getDevice( index );

Error: nnet.internal.cnn.util.isGPUCompatible (line 9)

if(iCanUsePCT() && parallel.gpu.GPUDevice.isAvailable())

Error: nnet.internal.cnn.util.GPUShouldBeUsed (line 17)

tf = nnet.internal.cnn.util.isGPUCompatible();

Error: vision.internal.cnn.trainNetwork>iSetupExecutionEnvironment (line 134)

GPUShouldBeUsed = nnet.internal.cnn.util.GPUShouldBeUsed( ...

Error: vision.internal.cnn.trainNetwork (line 43)

executionSettings = iSetupExecutionEnvironment( opts );

Error: fasterRCNNObjectDetector.trainRPN (line 175)

[rpn, info] = vision.internal.cnn.trainNetwork(dispatcher, layers, opts,

checkpointSaver);

Error: trainFasterRCNNObjectDetector (line 289)

[stage1Detector, rpn, info] = fasterRCNNObjectDetector.trainRPN(trainingData, rpn,

options(1), iStageOneParams(params), checkpointSaver);

Error: rcnn_car_object_detection (line 84)

detector = trainFasterRCNNObjectDetector(trainData, layers, options, ...

Answer 1

[Vidip]

I understand that you are getting the following error: error using parallel.gpu.CUDADevice.hBuild, with a CUDA runtime error of CUDA_ERROR_UNKNOWN. Try setting the "high performance NVIDIA Processor" option for MATLAB on NVIDIA control panel. You can find it at 3D Settings -> Manage 3D Settings -> Select a program to customize. This should alleviate the CUDA errors.

For more information regarding CUDA errors while training a network on GPU, you can refer to the below MATLAB Answers thread-

• https://in.mathworks.com/matlabcentral/answers/715848-why-do-i-get-cuda-execution-errors-when-training-my-network-on-a-gpu