Machine Learning Image Classifier

MachineLearning-image-classifier provides Jupyter notebooks demonstrating image classification using two approaches:

From scratch (from-scratch.ipynb) — train a CNN model from the ground up on a custom dataset
Transfer learning (from-pretrained.ipynb) — fine-tune a pre-trained model for new categories

Dataset Structure

.
├── from-pretrained.ipynb
├── from-scratch.ipynb
├── single-test/
│   └── unknow/
├── train-data/
│   ├── cat/
│   ├── dog/
│   └── caduckrbord/
└── validation-data/
    ├── cat/
    ├── dog/
    └── duck/

Setup

python -m pip uninstall numpy
python -m pip install numpy==1.23.1

Links

MachineLearning-image-classifier GitHub repository