Building a Satellite Image Classification Model with Residual Neural Network

EasyChair Preprint 13930

Abstract

Satellite image classification plays a crucial role in various fields such as agriculture, urban planning, and environmental monitoring. Accurate classification of satellite images helps in extracting valuable information and making informed decisions. In recent years, deep learning models, particularly Residual Neural Networks (ResNet), have shown remarkable performance in image classification tasks. This abstract presents an overview of building a satellite image classification model using a ResNet architecture.

The process begins with data preparation, including gathering a dataset of satellite images and preprocessing the data through resizing, cropping, and normalization. The dataset is then divided into training, validation, and testing sets to facilitate model development and evaluation.

The Residual Neural Network is constructed by defining its architecture, which consists of convolutional layers with residual blocks, pooling layers, dense layers, and an output layer. The model is compiled with an appropriate loss function and optimizer. The training process involves setting up parameters such as batch size, number of epochs, and learning rate. The model is trained on the training set, and the training progress is monitored using evaluation metrics and visualizations.

Keyphrases: Model Deployment, ResNet, TensorFlow, compilation, loss function, model training, testing

@booklet{EasyChair:13930,
  author    = {Kayode Sheriffdeen and Samon Daniel},
  title     = {Building a Satellite Image Classification Model with Residual Neural Network},
  howpublished = {EasyChair Preprint 13930},
  year      = {EasyChair, 2024}}