A simple project designed to classify photos by various metadata tags, as well as to use OpenCV library to detect plagiarism among photos.
Check out video1 and video2 to have a quick overview over this tool.
First of all, I will tell you why I needed it so bad. Whenever I travel (especially hiking), I take photos both with my old Canon EOS 7D, and my mobile phone(it can easily replace a DSLR in some situations). Sorting those photos in meaningfully named folders takes a lot of my time and energy, since they are somehow strangely named (etc. 12319028391819SDSA.jpg), and eventhough I switched to "Show big tiles" in my folder preview mode, I cannot effortlessly differentiate which photos belong to which device. Same story goes for classifying in Adobe Lightroom (entering letters A, B, C in Caption field), and for various other data.
Why I added a plagiarism tool?
I thought that the first part of this project was a bit "to easy" to satisfy this course project, so I wanted a further challenge. Even though OpenCV is a popular library with great documentation and tutorials in Python and C++, it was a tough journey to put it all in a meaningful context(tons of documentation to read). The deeper I dived into Clojure, the more I realised I was naming my variables and functions in a "boilerplatecodegenerator" manner. Even though I found it easier this way, I will try to avoid this practice in my next projects.
That's it. Hope you will like it. Remember, it is my first Clojure project! :)
Not much. You will need Leiningen installed.
To start the application, open the terminal and run:
lein run
When the program runs, the user has an option to choose source directory of the photos, as well as
the target directory where photos will be classified by Caption, Artist or Camera make model.
Furthermore, user has an option to choose whether to run a plagiarism check. If one decides
to run a plagiarism check, that will result in creating a unique report of the matched pairs that
can be found in reports folder in project root directory.
The business logic for extracting metadata from photos was completed with the help of MetadataExtractor library.
The business logic for plagiarism detection was developed using OpenCV template matching functions. The method that was used to compare these photos is called TM-CCOEFF-NORMED - correlation coefficient method. TM_CCOEFF method is simply used to make the template and image zero mean and make the dark parts of the image negative values and the bright parts of the image positive values. This means that when bright parts of the template and image overlap you'll get a positive value in the dot product, as well as when dark parts overlap with dark parts.
The UI of this project was developed using Seesaw library.
This project was developed for the purpose of the course "Tools and Methods of Soft. Eng. and AI" at Master studies at University of Belgrade, Faculty of Organizational Sciences.
Midje library has been used for unit testing. I have tried to implement TDD (Test Driven Development) approach during the process of creating this software tool, by simultaneously creating features and testing them.
I did some quick-benching here also.
I compared performance of map and pmap functions in this project.
The conclusion is
that pmap outperforms map in the context where those were faced.
The goal of this project was getting myself familiar with Clojure programming language, as well as it's libraries and key strength. I did some serious brainstorming on which real-life problems could my program solve. Then, I did research on libraries that could be integrated with Clojure code and also help me on my way to create a useful program. Then - I got my hands dirty. The development of this project took way more than I thought it would take (my first estimation was 30 hours), but I am glad that it managed to come up as a useful tool. I am happy that I have found out this cool & powerful programming language. I most sincerely hope that I will find a way to furthermore enhance my knowledge of Clojure, especially in the field of AI, as well as to combine it with Android mobile development (somehow).
[1] Introduction to OpenCV-Python tutorials (https://docs.opencv.org/3.4/d0/de3/tutorial_py_intro.html)
[2] Howse J., 2013, OpenCV Computer Vision with Python, Packt publishing
[3] Kaehler A., 2017, Learning OpenCV 3: Computer Vision in C++ with the OpenCV Library, O'Rilley
[4] https://github.com/drewnoakes/metadata-extractor
[5] Seesaw REPL Tutorial (https://ericnormand.me/article/seesaw-repl-tutorial)
[6] https://cljdoc.org/d/seesaw/seesaw/1.5.0/api/seesaw.core#button-group
[7] https://github.com/marick/Midje
Copyright © 2022 FIXME
This program and the accompanying materials are made available under the terms of the Eclipse Public License 2.0 which is available at http://www.eclipse.org/legal/epl-2.0.
This Source Code may also be made available under the following Secondary Licenses when the conditions for such availability set forth in the Eclipse Public License, v. 2.0 are satisfied: GNU General Public License as published by the Free Software Foundation, either version 2 of the License, or (at your option) any later version, with the GNU Classpath Exception which is available at https://www.gnu.org/software/classpath/license.html.