/Image-Classifier-Project

Deep learning project to classify images using PyTorch, neural networks, and transfer learning with GPU.

Primary LanguageJupyter Notebook

ImageClassifierProject

Deep Learning Project to classify Images using PyTorch, Neural nets, GD, and transfer learning with GPU

Part 1: ImageClassifierProject.ipynb

Contents:

  • Introduce, load and preprocess image dataset
  • Building and training the classifier
  • Function to build a Neural Network and use pretrained model (e.g. vgg16, densenet121).
  • Batch Gradient Descent to optimize classifier using GPU Power
  • Testing the trained network, saving and loading checkpoints
  • Inference for classification
  • Function to predict class of an Image
  • Sanity checking

Part 2: train.py, predict.py

  • Develop an AI application which can be used from the command line

Function 1: train.py

  • Trains a neural network classifier.

Function 2: predict.py

  • Predicts likelihood that Image belongs to a certain category trained with train.py

Other files:

  • tm_fun.py: some helper functions for part 2
  • cat_to_name.json: Consists of the mapping of categories to flower names
  • workspace-utils.py: Keeps workspace awake if GPU takes longer...