Vijaybedage/ML-project — reverse-engineered prompt

Build me a simple animal image classification project in Python that can recognize 15 animals from a photo, like bear, bird, cat, cow, deer, dog, dolphin, elephant, giraffe, horse, kangaroo, lion, panda, tiger, and zebra. I want it to use transfer learning with MobileNetV2, include a training script for a dataset organized by animal folders, and save the trained model so I can reuse it later.

Please also make a small Streamlit app where I can upload a jpg, jpeg, or png image and get a prediction right away with a confidence score, top 5 guesses shown in a chart, and a short fun fact for the predicted animal. It should feel clean and easy to use. Include the usual evaluation outputs like training history, confusion matrix, class report, and dataset distribution, plus a simple single image prediction script from the command line. Use sensible defaults like 224 image size, augmentation, early stopping, and an 80 20 train validation split. If anything is missing, look up current docs online and make it work end to end.