The integration of artificial intelligence and machine learning through computer vision (CV) technologies has gained widespread acceptance for the evaluation of traffic safety. While these technologies offer substantial advantages, they are accompanied by constraints and cost implications. This paper introduces a pragmatic framework for selecting a cost-efficient CV system for assessing traffic safety at intersections. The approach is designed based on Weighting Rating and Calculating (WRC) model and involves evaluating the tasks across all stages of the projects encompassing video collection, pre-processing, video analytics, and traffic safety analysis to aid in informed technology selection. The framework was applied to real-world projects, with resultant observations and outcomes discussed. The projects aimed to accurately identify conflict zones within intersections, where complex interactions and behaviours occur among categorized vehicles and vulnerable road users. Future enhancements to the framework and the software are also proposed. This paper intends to offer transport authorities and consultancy firms a structured methodology for choosing cost-effective technologies that align with their intersection safety assessment needs.