In order to meet the needs of smart munitions to respond to complex confrontation combat, with the upgrading of key components such as high-performance detectors and information processors, engineering applications of target recognition technology based on deep learning have emerged, aiming at the future development trend of a new generation of photoelectric seekers, the multi-mode composite guidancing system, integrated miniaturization, and intelligent practical applications and low-cost design were expounded. The advantages of using multi-mode composite photoelectric seekers such as visible light, infrared and laser were analyzed to extend the range of missile terminal guidance, adapt to all-weather all-day work, and improve anti-jamming capabilities. It is pointed out that the overall integration of the optical machine and the integration of the information processing system are the key technologies to solve the miniaturization of the composite seeker. The intelligent characteristics of seeker such as automatic target recognition and multi-projectile coordination are emphasized. Proposals are put forward to realize the low-cost engineering application of photoelectric seeker by using technologies such as miniaturized common aperture multi-mode photoelectric composite detection, integrated information processing, low-cost technical system, low-cost manufacturing technology and low-cost combat support.