Wild bees and wasps are vital to ecosystems, yet large-scale monitoring of individual insects as well as their habitats and behaviors requires expert knowledge and is very labour-intensive. Recent advances in computer vision offer both methodological solutions and practical applications for these tasks. This review systematically surveys state-of-the-art literature (2020-2025) on computer-vision-based monitoring of wild bees and wasps. We compare the primary monitoring tasks, such as individual detection and classification, habitat observation, and assessment of insect behavior, and analyze how specific computer vision techniques contribute to each. By examining the datasets used in the reviewed studies, we further categorize dataset types and collection strategies to inform future image acquisition. In addition, we draw insights from widely used public resources (e.g., iNaturalist, Observation) regarding their strengths and limitations for this domain. We then examine how hardware and software are integrated in these studies, and review released repositories, existing applications, and the design of more complex, multifunctional monitoring stations. These analyses provide guidance for benchmarking and future deployment with existing datasets and monitoring tools. We eventually propose the design of building more comprehensive and efficient automatic monitoring systems for wild bees and wasps.