An in-vehicle face image detection apparatus operates as follows. A projector projects light. A capture controller settles a capturing condition to capture a predetermined region including a face of an occupant. A capturing portion captures the predetermined region based on the capturing condition. An environment light detection portion detects environment light radiated to the vehicle or occupant. An operation mode determination portion determines a first mode where intensity of the environment light is greater than a threshold or a second mode where the intensity is not greater than the threshold. A light emission pattern setup portion settles a light emission pattern for a light source of the projector based on the operation mode. A projector controller controls the projector to activate the light source based on the light emission pattern.