Mihir-Arjun/Big-5-Image-Recognition

This is a project that makes use of a deep CNN in order to classify animal images into the big 5 South African Animals # Big Five Image Recognition Project A project by Mihir Arjun ## Table of Contents - Image Recognition for Classifying the Big Five Animals of South Africa - Data Preparation and Cleaning - Loading and Visualising the Data - Data Scaling - Splitting the Data - Building the Deep Learning Model - Training the Model - Plotting Model Performance - Evaluating the Model - Testing the Model - Summary ## Image Recognition for Classifying the Big Five Animals of South Africa This project focuses on classifying images of the Big Five animals of South Africa using a Convolutional Neural Network (CNN) built with TensorFlow and Keras. The Big Five includes buffalo, elephant, leopard, lion, and rhino. The guide provides a step-by-step approach to building, training, and deploying a CNN for this classification task. ## Data Preparation and Cleaning - **Image Validation:** Validate each image using OpenCV to ensure it is not corrupted. - **Format Verification:** Verify the image file type against supported formats. - **Removal of Invalid Images:** Remove corrupted or unsupported images to maintain dataset integrity. ## Loading and Visualising the Data Use TensorFlow's `image_dataset_from_directory` utility to load images and automatically infer labels based on subdirectory names. Visualise a few sample images to verify correct loading and labeling. ## Data Scaling - **Normalising Pixel Values:** Normalise pixel values to a range between 0 and 1 by dividing by 255.0. - **Verifying Scaling:** Fetch and visualise a batch of images post-normalisation to ensure proper scaling. ## Splitting the Data Split the dataset into: 1. **Training Set (70%)**: For model training. 2. **Validation Set (20%)**: For hyperparameter tuning and monitoring. 3. **Test Set (10%)**: For final performance evaluation. ## Building the Deep Learning Model - **Initialising the Model:** Create a Sequential model using Keras. - **Adding Convolutional and Pooling Layers:** Extract spatial features and …