Upcoming large area surveys, such as Euclid and LSST, will produce high quality imaging data for billions of galaxies. Such a large volume of data cannot be hoped to be analysed by hand and so we need new, fast and reliable methods to identify the rare objects in these data sets. The NEP field offers an interesting test environment to develop these new techniques, with its wealth of high quality, multi-wavelength data, while being small enough to allow human validation of the results.   Galaxies undergoing a merger are uncommon but underpin our current understanding of how galaxies grow and evolve. The tidal forces during these events move the stars, dust and gas and can change the morphology of the merged galaxy. This movement of material can also trigger extreme events, such as triggering active galactic nuclei or periods of intense star formation. The merger fractions is also dependent on the environment in which a galaxy lies.  In this presentation, I will discuss deep-learning techniques being applied to the NEP field to identify merging galaxies. This will explore how the existing high quality data in other fields can be used to develop and train a neural network to be applied in the NEP field. I will also discuss how a galaxy’s place in the field, group or cluster environments may influence the merger fraction using these merger identifications. The speed of classification will also be discussed in the context of the upcoming, all-sky surveys along with current limitations of merger identification.