Marine debris interferes with the movement of ships and wildlife and, in turn, harms humans. This study detects nighttime port garbage to establish a sound marine ecology and provide maintenance information to environmental protection units. Since the port only relies on the weak street lights on the shore as a lighting source at night, it is generally a dim scene. Therefore, this study uses low-brightness image processing methods to visually capture and analyze the garbage in the port and artificial intelligence algorithms to analyze the garbage distribution. Eventually, this study obtained the night with low-brightness image processing and the night and day modeling without low-brightness image processing, with average error rates of 5%, 72%, and 246%, respectively. Therefore, this study uses low-brightness image processing methods and artificial intelligence algorithms for nighttime harbor garbage sensing, which has a good sensing effect in monitoring the port's water quality.