trainFastR​CNNObjectD​etector - Unable to find any region proposals to use as positive training samples

Dear collegues,

I already tried everything to train a fast RCNN Object detector with trainFastRCNNObjectDetector (downsample my training images, larger bounding boxes, smaller bounding boxes, lower PositiveOverlapRange) but I always get the same error:

"Error using rcnnObjectDetector.imageRegressionInfoFromGroundTruthDatastore (line 1173)

Unable to find any region proposals to use as positive training samples. Lower the first value of PositiveOverlapRange to increase the number of positive region proposals."

• I have one class;
• My images are grayscale (862x662)
• My images are .tif
• The table as input for training is generated through Video Labeler (using ground truth object for objectDetectorTrainingData(gTruth,'SamplingFactor',1);
• The layers as input for training are (from the example https://www.mathworks.com/help/vision/ref/trainfastrcnnobjectdetector.html#namevaluepairarguments):

fastRCNNLayers = [

imageInputLayer([32 32 1],"Name","imageinput")

convolution2dLayer([3 3],32,"Name","conv_1","Padding",[1 1 1 1])

reluLayer("Name","relu_1")

convolution2dLayer([3 3],32,"Name","conv_2","Padding",[1 1 1 1])

reluLayer("Name","relu_2")

maxPooling2dLayer([2 2],"Name","maxpool","Stride",[2 2])

fullyConnectedLayer(64,"Name","fc_1")

reluLayer("Name","relu_3")

fullyConnectedLayer(2,"Name","fc_2")

softmaxLayer("Name","softmax")

classificationLayer("Name","classoutput")];

What am I doing wrong?? Please any help!!