metadata

tags:
  - image-classification
  - tensorflow
  - keras
  - computer-vision
  - animal-recognition
license: apache-2.0
library_name: keras
language: en
datasets:
  - custom-dataset
metrics:
  - accuracy
model_creator: AEROVERSE
model_type: Custom CNN
num_classes: 10
training_data: Custom animal dataset
validation_split: 20%
epochs: 500
early_stopping: patience=5, restore_best_weights=True
dropout: 0.5
optimizer: Adam
loss: sparse_categorical_crossentropy
input_shape: (256, 256, 3)
output_activation: softmax
checkpoint: best_model.weights.h5
pipeline_tag: image-classification

Animal Recognition Model

Model Overview

This model is designed to classify images of animals into predefined categories. It uses a ResNet50V2 base model and has been trained on a custom dataset.

Classes

The model was trained on the following classes:

cat
dog
horse
lion
tiger
elephant

Usage

Load the model using TensorFlow/Keras.
Preprocess the input image to a size of 256x256 and normalize it.
Pass the preprocessed image to the model for prediction.

from keras.models import load_model
import numpy as np
from tensorflow.keras.utils import load_img, img_to_array

def predict_image(image_path, model):
    img = load_img(image_path, target_size=(256, 256))
    img_array = img_to_array(img) / 255.0
    img_array = np.expand_dims(img_array, axis=0)
    prediction = model.predict(img_array)
    return np.argmax(prediction, axis=1)

model = load_model('best_model.weights.h5')
predicted_class = predict_image('image.jpg', model)
print(f"Predicted class: {predicted_class}")

Training Details

Base Model: ResNet50V2 (pretrained on ImageNet)
Dataset: Custom animal dataset
Optimizer: Adam
Loss Function: Sparse Categorical Crossentropy
Metrics: Accuracy
Augmentation: Applied during training

Model Performance

Will be updated soon