The rapid advancements in information technology and artificial intelligence (AI) have fueled exponential data growth and an unprecedented demand for computational capabilities. However, the pace of computational power enhancement through integrated circuit technology advancements has struggled to keep up with the soaring needs of AI. Furthermore, traditional electronic computing systems, constrained by the Von Neumann Architecture, struggle to meet the stringent requirements of speed and power consumption. Optical computing systems emerge as promising solutions, addressing the computational limitations and power challenges faced by their electronic counterparts. At the heart of optical computing lie optical neural networks, realized through optical hardware, which inherently facilitate mathematical operations such as convolution, differentiation, and integration in a physical manner. Leveraging their inherent advantages of high parallelism, wide bandwidth, blazing speeds, and minimal power consumption, optical neural networks offer a viable path to alleviate the computational and power constraints hindering AI. Consequently, they hold significant potential for applications spanning image recognition, edge detection, voice recognition, and beyond.