Neural Image Classification

This repository contains a simple animal image classification example using TensorFlow and Keras. The code snippets demonstrate how to:

- Load and preprocess images
- Define a neural network model
- Compile and train the model
- Evaluate the model's performance
- Make predictions
- Capture an image from a laptop camera
- Preprocess and classify a new image

Model

The neural network model is defined below and consists of several layers:

- Conv2D (16 filters, kernel size 3x3, activation='relu')
- MaxPooling2D (pool size 2x2)
- Conv2D (32 filters, kernel size 3x3, activation='relu')
- MaxPooling2D (pool size 2x2)
- Flatten
- Dense (64 units, activation='relu')
- Dropout (0.5)
- Dense (3 units, activation='softmax')

The model is compiled with the Adam optimizer and categorical cross-entropy loss function.

Training

The model is trained on the training set using the fit method. The training process is repeated for 10 epochs with a validation split of 0.2.

Evaluation

The model's performance is evaluated on the validation set using the evaluate method. The loss and accuracy are printed to the console.

Prediction

The model makes predictions on a new image captured from the laptop camera using the predict method. The predicted label is printed to the console.

Requirements

- TensorFlow 2.x
- Keras 2.x
- NumPy
- SciPy
- Pillow
- OpenCV

Usage

To run the code, simply execute the Python scripts in the following order:

1.  train the model
2.  capture an image from the laptop camera
3.  preprocess and classify the captured image

-NOTE: use a larger and more diverse dataset to improve the model's accuracy.

# Some parts of the codebase are adapted from "https://www.tensorflow.org/guide/keras/sequential_model", 
"https://pyimagesearch.com/2018/12/31/keras-conv2d-and-convolutional-layers/",
"https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html"


Detail explanation: https://medium.com/@omarjbt/designing-a-custom-neural-network-model-for-image-classification-using-your-own-images-as-the-089d36b13658