image_classification

This repository contains convolutional deep learning neural networks projects.

Projects

• Dog Breed Classifier: Build a convolutional neural network, creating a convolutional network from scratch as well as using transfer learning, with PyTorch in ordert to classify any image (even an image of a face) as a specific dog breed.
• Image classification_with_tensor_flow: Build a convlutional neural network with Tensor flow that classifies up to 10 different classes.

Installation

1. Clone the repository
2. Download the latest version of miniconda that matches your system.
3. Create and Activate a conda environment.

• Linux or Mac:

 conda create -n "your_folder_name" python=3.6
 source activate "your_env_name"

• Windows:

conda create --name "your_folder_name" python=3.6
activate "your_env_name"

4. From the conda environment, install PyTorch and torchvision.

• Linux or Mac:

conda install pytorch torchvision -c pytorch

• Windows:

conda install pytorch -c pytorch
pip install torchvision

5. Install other dependencies using pip (See requirements text file including opencv and jupyter notebook).

pip install -r requirements.txt

6. In case you want to run Image_classification_with_tensor_flow as well you need to install tensorflow.

pip install tensorflow