Image manipulations that violate the rules of good scientific practice are a serious problem in many scientific fields. Especially if erroneous or deliberately falsified images go undetected, there is a risk that research based on them will produce incorrect results.
Detecting image manipulation is by no means trivial, and numerous research teams worldwide are currently working on developing algorithms for the automated detection of manipulations. Comparing and testing the efficiency of such algorithms requires comparable test data. The Image Integrity Database of the HEADT Centre collects images from retracted publications that have encountered problems with images. This comprehensive data pool makes it possible to test algorithms for manipulation detection on images from actual research.