YOLO v3 - Error message while setup of training

Hello,

i am trying to set up the training for a YOLO v3 - object detector. Everything is pretty much copied from the example website

https://de.mathworks.com/help/vision/ug/object-detection-using-yolo-v3-deep-learning.html

Now, at the line:

[gradients, state, lossInfo] = dlfeval(@modelGradients, yolov3Detector, XTrain, YTrain,
penaltyThreshold);

I get the following error message, and i do not know how to fix this:

Check for incorrect argument data type or missing argument in call to function 'forward'.
Error in modelGradients (line 8)
dlYPred = forward(dlnetDiscriminator, dlX, dlT);

Does anything have an idea of how to fix this?

Thanks already in advance!

Answer 1

[Maneet Kaur Bagga]

Hi Osman,

As per my understanding of the question, "dlfeval" is used as a handler for the modelGradients function which takes the "yolov3ObjectDetector" object, a mini-batch of input data XTrain with corresponding ground truth boxes YTrain, the specified penalty threshold as input arguments and returns the gradients of the loss with respect to the learnable parameters in "yolov3ObjectDetector", the corresponding mini-batch loss information, and the state of the current batch.

The error is due to the target data "dlT" paased as an input argument in the "forward" function, the forward function expects the input parameters to be the detector object and the training dataset and returns the output features of the network during the training phase. Removing the train data parameter “dlT” will resolve the error occured.

features = forward(detector,dlX);

Please refer to the following MATLAB Documentation for better understanding of the forward function:

https://www.mathworks.com/help/vision/ref/yolov3objectdetector.forward.html

Hope this helps!

Thank You

Maneet Bagga