An occupant detection device includes: a face area detecting unit for executing a process of detecting a plurality of face areas corresponding to a plurality of faces in an image captured by a camera for capturing an image of a vehicle interior; a first determination unit for determining whether or not, among a plurality of seat areas corresponding to a plurality of seats in the captured image, there is a seat area including two or more face areas; a second determination unit for determining whether or not a duration time of a state where there is a seat area including two or more face areas has exceeded a reference time; and a warning output control unit for executing control for outputting a warning when the duration time exceeds the reference time.