anandchauhan21/pomegranate_disease_prediction ? reverse-engineered prompt

Build me a pomegranate disease prediction app from image data, using a CNN model in TensorFlow.

I want two parts, one model that looks at a pomegranate fruit photo and tells me if it has a disease, and another model that checks pomegranate leaves and identifies the leaf disease too. After the prediction, show the disease name and a simple solution or treatment suggestion for that disease in plain language.

Please make it work in notebook form like a Colab friendly project, with one notebook for training the models and another for testing predictions from new images. Use OpenCV where it helps for image loading and preprocessing, and keep the workflow easy to follow so I can retrain it with my own dataset later. If you need to check the latest TensorFlow or OpenCV docs online, go ahead.