Social inequalities are being reinforced and propagated by the economic operations of the tech industry, specifically in the ways that data is scraped and trained for AI technologies. The blind trust that is placed in software to provide "objective" judgements creates problematic divisions between those who can shape knowledge, and those who are tracked, silenced, and exploited to develop technologies for First World users. This problem, data colonialism, recalls a European colonial tradition, wherein a scientific, 'objectivist' rationale was provided for the colonization of 'other' groups. To address data colonialism, a data literacy project was implemented, to educate users on how "truth" is created using data, and to provide new possibilities for self-representation for migrants, whose stories are not heard in the mainstream media. For this project, we decided to focus on how experiences of migrants could be incorporated into a computer vision dataset. A webpage allows users to upload pictures in a database, tag them with their own perception and subsequently compare their vision with that of a computer, for which an image classification system was implemented. Furthermore, the tags used and their relationships are drawn into a network graph, that represents a semantic map of user experiences. Technologies used include React.js, Material-UI, FastAPI, Python with libraries like Keras and SQLalchemy.