Pneumonia Prediction with EfficientNetB0

Overview

This project focuses on predicting pneumonia using a machine learning model trained on the EfficientNetB0 architecture. With a solid 79% accuracy, the model showcases efficient feature extraction, making it a valuable tool for medical diagnostics.

Features

• Efficient Model: Utilizes EfficientNetB0 for feature extraction.
• Accuracy: Achieves an accuracy of 79% on the provided dataset.
• Data Source: Kaggle dataset forms the foundation for training and testing.

Objective

Develop a machine learning model for pneumonia prediction using chest X-ray images and transfer learning.

Data

• Source: Kaggle dataset- Chest X-Ray Images (Pneumonia)
• Preprocessing: Divided the training images into batches of 32 for better results in training

Model

• Architecture: EfficientNetB0 as the feature extraction layer and a Dense output layer
• Training: Trainable parameters are 2049/ 23566849
  • Optimizer: Adam
  • Loss Function: binary_crossentropy

Evaluation

• Accuracy: 79%

Results

The model achieved an accuracy of 79% in pneumonia prediction.