Overview / Usage

Using learning transfer to create a model classifier on the flower 102 dataset. Final result is a CLI application with the ability to allow user to set model training parameters, base architecture etc.

Methodology / Approach

1.)Preprocess dataset for training: Color normalization and center cropping to 256x256

2.)Load backbone model: Loading the base model(resnet152,vgg16 etc.)

3.)Building classification layer: Building a classification layer atop the backbone model

4.)Training

5.)Predictions

Technologies Used

1.)PyTorch

2.) Nvidia CUDA

3.)Linux/BASH

Repository

https://github.com/AlexFierro9/Image-Classifier