Novel object recognition (NOR) task is amongst the most commonly used behavioural test for investigating hippocampal and parahippocampal related memory in rodents. It is based on an animal’s natural tendency for increased exploration of novel objects compared to familiar ones, and does not require food or water restrictions for motivation. However, its implementation is often challenging due to high variability associated with exploratory behaviours. Here, we present an unsupervised novel object recognition task implemented in a homecage environment (smart-Kage; Fig. a). During the 24-hour sampling phase, a mouse was presented with two distinct novel patterns (‘objects’), with which the mouse could directly interact by touch, vision, and smell. The sampling phase was followed by either 24 hours (n=20 mice) or 7 days (n= 19 mice) inter-trial intervals (Fig. b). During the test phase, mice were presented with two objects: a duplicate of a previously encountered object and another one that was unfamiliar. The object presentation was randomised to balance the object identities and location within each test cohort. First, we showed that on average there was no statistically significant bias in the preference for object identity or location during the sampling and test phases (Fig. c-d). Next, we showed that mice were able to recall familiar objects after an interval of 24 hours and 7 days (Fig. e) and our systematic methodology minimised behavioural variability between individuals. We expect that the smart-Kage automated NOR task will enable higher throughput and more standardised memory testing in mice.