The quality of underwater photographs has great importance in the diverse field of oceanic engineering, particularly for robotic exploration and marine biological applications. By combining high-level semantic characteristics with low-level picture qualities, this research presents a unique underwater image quality assessment (UIQA) method that overcomes existing shortcomings. Incomplete evaluations of picture quality are produced by traditional UIQA techniques, which usually concentrate on either high-level semantic information or low-level attributes like brightness and contrast. By including essential elements of underwater picture quality evaluation, the suggested technique closes this gap. The method extracts high-level semantic information and low-level perceptual elements from underwater photos by using the BRIQUE and NIQE metrics to calculate Mean Opinion Scores (MOS). The estimated MOS values are utilized to train the model, while SRCC, KRCC, and PLCC metrics are used to evaluate performance by comparing anticipated and real MOS. The strategy sets a new standard for accuracy and dependability in underwater image evaluation, since experimental findings show that it far improves current UIQA procedures.