This project implements an image classifier using a deep learning model trained on the 102 Category Flower Dataset. The goal is to develop a neural network capable of classifying images into different flower species.
Initially, training was performed on provided Workspace, but due to several issues, the training process was moved to Kaggle for better computational resources and efficiency. The implementation is inspired by Python resources, Udacity course materials, and various GitHub repositories.
The project takes inspiration and guidance from various sources like google, gitHub,resources provided in this program,etc
- Uses a pretrained model for feature extraction.
- Implements transfer learning for better accuracy.
- Includes data augmentation for improved generalization.
- Trains and evaluates the model on Kaggle GPUs.
- Exports the trained model for inference.
To run the project, install the required dependencies:
pip install torch torchvision numpy matplotlib pandas- Load the dataset and preprocess the images.
- Define a convolutional neural network (CNN) using a pretrained model.
- Train the model using PyTorch.
- Evaluate the model on test data.
- Achieved more than 70% accuracy on the validation set.
- Model generalizes well to unseen images.
- Fine-tune the model for better accuracy.
- Deploy as a web app for real-time predictions.