With large telescopes, modern adaptive optics systems, and precision coronagraphs to block stellar light, we can take extremely high resolution images of nearby stars: we can image faint companions that emit a million times less light than their host stars, at a separation of less than one thirty-thousandth of a degree. Yet the exoplanet community have only directly imaged roughly 10 exoplanets in the last two decades, during which time several thousands of transiting planets have been discovered. The problem with direct imaging is that telescope time is expensive and stars must be surveyed individually, and that massive planets on wide orbits are rare. Even when a source near a nearby star is detected, we face the problem of differentiating a faint planet at the distance of the target star, and a faint background star that is much much further from Earth. In this talk I’ll discuss my attempts to directly detect an exoplanet, and some of the observational and data processing techniques we use to find these very faint objects. I’ll also explain what we know about the directly imaged exoplanet population as a whole, and why no-one actually knows the exact number of exoplanets that we have directly imaged. I’ll then tell you about a project I’ve been working on more recently, where we use some of the same techniques with the explicit goal of finding background stars, so as to validate transiting planets found by the TESS mission. Ultimately it will take many different detection methods to fully understand the diversity of planetary systems, and how they form and evolve.