simple-image-classification

This repository contains:

• simple_image_classification.ipynb: A Jupyter notebook to perform basic image classification and display the results in order to test adversarial images
• simple_video_classification.py: Python script to capture live video feed and perform image classification
• images folder containing sample images and adversarial patches (developed by engineers at www.roke.co.uk)

Setting up your environment

Instructions for setting up your enviornment to run the code are in GETTING_STARTED.md

Running the video classifier

From within an Anaconda Command Prompt, ensure that you are within the correct environment (see GETTING_STARTED.md)

conda activate simple-image-classification

Navigate to the folder containing the clone of this repository and type:

python simple_video_classificattion.py

A window like this should display:

Resize the window and ensure that it is the primary focus (click on it).

Point your web camera in the direction of the scene that you would like to capture and press c on the keyboard. You should see something like this:

Press c each time you want to update the scene being classified.

Press q to quit.

(You cometimes need to press c or q a couple of times if the thread us busy)

Running the image classifier

From within an Anaconda Command Prompt, ensure that you are within the correct environment (see GETTING_STARTED.md)

conda activate simple-image-classification

Navigate to the folder containing the clone of this repository and type:

jupyter notebook

Here's a good introduction to Jupyter notebooks

When running the Jupyter notebooks, the correct kernel Python (simple-image-classification) (created in the one time step previously). This is selectable via a drop down at the top right of the Jupyter notebook interface.