Nvidia DIGITS Caffe model using VGG16 SSD 300 300 model and deploy prototxt


I am trying to use Digits to train SSD object detection task for 38 classes (39 including background).

I have deploy.prototxt and .caffemodel files with.
I have data set with varying sizes. In DIGITS i gave resize to 300*300.

dataset is read properly.
But when I put my deploy.prototxt and pretrained model .caffemodel files.
I am getting error like:

error: check failed: num_priors_ * num_classes_ == bottom[1]->channels() (340548 vs. 183372) number of priors must match number of confidence predictions.

Could you please let me know what should be modified in deploy.prototxt file .

Thank you