Hi Dave, thank you for your note. Please cross verify if you work is significantly different from the CNN models built on this problem in my textbook Understanding Deep Learning. It is open to download at www.understandingdeeplearning.com . If they are different I will be happy to give you consent to publish on TDS.