In recent years, there has been substantial progress in measuring mixing patterns relevant for disease transmission with wearable sensors. The progressive miniaturization of the devices and their decreasing costs have allowed to scale-up large contact experiments up to hundreds of individuals across a variety of settings. Some examples include hospitals, schools, conferences and more recently, households. However, so far, the deployment of wearable proximity sensing technologies has been mostly limited to high developed countries in Europe or in the USA. On the other hand, social mixing patterns in low-resource settings remain an important yet understudied research topic because of the practical challenges in conducting survey research in such areas. Here, we report on some recent deployments of wearable proximity sensors in two different resource-poor settings in Africa. The proximity sensing technology used in these studies has been originally designed by the SocioPatterns (http://sociopatterns.org) collaboration consortium. It provides an affordable and fully-distributed sensing platform to measure high-resolution face-to-face proximity networks. We will present results of two pilot studies and two large-scale deployments of the SocioPatterns platform that took place between 2012 and 2018, showing that the collection of social mixing data in resource-poor settings is feasible and affordable using wearable proximity sensors. More specifically, the first large-scale deployment involved about 300 individuals belonging to 9 extended households in a rural area of coastal Kenya. The second large-scale deployment is an ongoing longitudinal study, planned to last until the end of 2018, that involves more than 700 individuals in about 100 households, located in two different sites of rural South Africa. The aim of the presentation will be to discuss the main challenges in the study design, the community engagement, the field deployment and the data collection, for these settings.