Study of the Adaptability of ResNet50 to new Images with Disparate Context

Overview

Our project utilizes Docker, JupyterLab, and standard Python to provide multiple avenues to run our project.

Prerequisites

• Docker Installed on Your Machine Get Docker
• Understanding of the Command Line and Docker

Setting Up Docker

Building Docker Image

Building the Docker image just requires running the build.sh script. This script will create an image name nb_dl_project and starts a container with this image named nb_dl_container.

./build.sh

Rebuilding the Docker Environment

Rebuilding the Docker container uses the rebuild.sh script. This script removes the existing container, rebuilds the image, and starts a new container with this rebuilt image.

./rebuild.sh

Using the Project

Accessing JupyterLab

After running build.sh or rebuild.sh, JupyterLab will be available at http://localhost:8888. The command line will provide a token for the session.

Restarting the Same Container

If you want to restart the container without making any changes to the Docker image, you can simply start the container again. You can do this with the following Docker command:

bash 
docker start -ai nb_dl_project

Running with Python

This project contains a .py file that the user can run with their local Python interpreter. This project was written targetting Python 3.9.7.

Command /usr/local/bin/python3 ./DeeplearningProject/test_proj.py

Hyper-Parameter Tuning with RayTune

Due to how RayTune runs its testing environment, hyper-parameter tuning requires that all relative paths be replaced with their absolute path for the program to run. The relative paths have been left in for users so that they can fill in with the path to this folder on their computer.