I-Jayswal/AfricanWildLifeDetection

The project aims to build an object classification model using a pretrained network to identify four types of wildlife in Africa: buffalos, elephants, rhinos, and zebras. It details the pipeline, metrics, results, and evaluation of the model's performance. # Wildlife Classification Project Overview This project involves building an object classification model to identify four wildlife species in Africa: buffalos, elephants, rhinos, and zebras. We leverage the VGG16 pretrained model, a deep convolutional neural network, to classify these animals. Data Augmentation Due to a limited number of training images, we use data augmentation techniques to increase the dataset size and reduce the risk of overfitting. Model Pretrained Model: VGG16, with weights from ImageNet, is used for this project. No additional trainable parameters are added to the pretrained model. Testing and Evaluation The model is evaluated using a test dataset, achieving an accuracy of 96%. Depending on the data quality, accuracy could potentially reach up to 98%. Pipeline and Results The pipeline, including data augmentation, model configuration, and evaluation, is detailed in the project documentation. Metrics and results demonstrate the model's effectiveness in wildlife classification.