A serious problem in biomedical science today is the proliferation of fraudulent and plagiarized scientific images in published articles and preprints. This has been talked about consistenly for the last several years, but few solutions have yet been proposed. OpenBioML and the Bioinformatics Research Network have joined forces to address this problem head on. Our collaborative effort is called "Scientific Image Search" (working title).
In this project, we will develop a dataset and model for detecting evidence of plagiarism in scientific images. We will also develop a web API and UI for analyzing new figures and exploring the database. If this project grows in funding and interest, we will also develop a model which can detect image manipulation in addition to plagiarism.
...