vincilbishop / image-recognition

image-recognition

Enhanced Computer Vision Using Automated Optimized Neural Network Image Pre-Processing

Reproduction Package

About

This repository was prepared for the IS&T Archiving Conference 2022

Description

Image recognition logic for testing.

The logic uses the following filters in 25% incremental increases:

• Contrast (0%, 25%, 50%, 75%, 100%)
• Background Removal (100%)
• Sharpen (0%, 25%, 50%, 75%, 100%)
• Despeckle (100%) The ideal code would run:

1. All possible combinations of these image alterations on a batch of photos in a folder we designate and save the new altered photos.
2. Use the google vision API to return the top classification results. I can separately run all the original photos for our baseline and write up an analysis of the end results. This should produce 990 alterations from which predictive modeling can be performed to focus on which more specific combinations would perform the best. For the photo sets, I was thinking of using a batch of aerial aircraft or simply vehicle photos since this could have significant impac

Methods

• Contrast: https://imagemagick.org/script/command-line-options.php#brightness-contrast
• Sharpen: https://imagemagick.org/script/command-line-options.php#sharpen
• Despeckle: https://imagemagick.org/script/command-line-options.php#despeckle
• Background Subtraction:

https://github.com/nadermx/backgroundremover

export JAVA_HOME=/Applications/Fiji.app/java/macosx/adoptopenjdk-8.jdk/jre/Contents/Home; /Applications/Fiji.app/Contents/MacOS/ImageJ-macosx --headless --console -macro ./subtract-background.java 'folder=./folder1 parameters=a.properties output=./samples/Output'

--headless --console -macro ./subtract-background.java 'folder=../folder1 parameters=a.properties output=../samples/Output'

/Applications/ImageJ.app/Contents/MacOS/ImageJ --headless --console -macro ./subtract-background.java 'folder=../folder1 parameters=a.properties output=../samples/Output' https://imagej.net/learn/headless

Libraries

The following libraries are candidates for use in the solution:

• https://github.com/aheckmann/gm
• https://www.npmjs.com/package/@google-cloud/vision

The following libraries are interesting:

• https://www.npmjs.com/package/imagej
• https://www.npmjs.com/package/opencv

Installation

To install imagemagick & graphicsmagick

$ brew install ImageJ
$ brew install imagemagick
$ # brew install graphicsmagick
$ yarn add imagemagick
$ yarn add @types/imagemagick --dev

Running the app

# development
$ npm run start

# watch mode
$ npm run start:dev

# production mode
$ npm run start:prod

Test

# unit tests
$ npm run test

# e2e tests
$ npm run test:e2e

# test coverage
$ npm run test:cov

Stay in touch

License

MIT licensed.

General Notes

https://blog.risingstack.com/opencv-tutorial-computer-vision-with-node-js/