Effective management of meal-related insulin dosing remains a critical challenge in diabetes care, often leadingto errors that can significantly impact glycemic control and longterm health outcomes. This study proposes an advancedsolution to address these challenges by integrating computer vision and machine learning technologies into diabetesmanagement. The research begins by thoroughly analysing the limitations of current insulin dosing practices, with a focuson identifying common errors and their consequences on patient health. Extensive data collection and user experienceanalysis are conducted to gain a comprehensive understanding of existing practices and inform the design of a moreaccurate, efficient system. The proposed system is designed to leverage image recognition to identify various food items andaccurately estimate their macronutrient content. Based on these estimations, the system calculates individualized insulindoses tailored to each user’s specific insulin sensitivity and needs. To ensure safety and minimize risks, robust error checkingmechanisms are incorporated, emphasizing accuracy and reliability in the insulin dosing process. This researchdemonstrates the potential of combining machine learning and computer vision to improve the precision and personalizationof insulin dosing. The proposed solution offers a promising advancement in diabetes care, with the potential to significantlyenhance patient quality of life by reducing dosing errors and optimizing glycemic control.