How to run our programs

There are two ways to run the notebooks in this folder:

1. Preferred:
   1. Unzip in this folder and upload it onto Google Drive
   2. Download data according to the section Download data below and put it in the same folder as the code
   3. Run the code. Authorization to access Google Drive is required when mounting.
2. Alternative:
   1. Download and unzip in this folder
   2. Create a conda environment according to environment.yml
   3. Download data according to the section Download data below and put it in the same folder as the code
   4. Remove all Google Drive mount code cells
   5. Adjust all the paths to the path where data is stored, e.g:

      data_dir = pathlib.Path('/content/gdrive/MyDrive/CS4243/data')

   6. Run the code

This folder consists of 4 notebooks:

• project-resnet.ipynb: Data preprocessing and transfer learning/finetuning of ResNet50V2.
• project-inception.ipynb: Transfer learning of Inception V3 (data pipeline is cut-and-paste from project-resnet.ipynb)
• project-simplecnn.ipynb: Train a simple 3-layer CNN using same data pipeline as project-resnet.ipynb
• visualization.ipynb: Visualize and inspect model deficiencies.

Trained models are available from https://drive.google.com/drive/folders/1lKvco8M4nsq1ChyirGbFB1TBdHyhjUHt?usp=sharing

Download data

Data can be downloaded from https://drive.google.com/drive/folders/1n7yLSLjhXeUHMOcfPb2Nq57xVh3RjgiZ?usp=sharing

Input images are stored in the data subdirectory, classfied into normal, threat and carrying to utilize Tensorflow in-built preprocessing capability.