Surveillance cameras are widely utilized as a resource to better control home environments such as monitoring children, the elderly, pets, and safety, among other applications. However, as it relates categorically to visual information, people who are visually impaired or blind are omitted due to the barriers that result from accessing information predicated on visual content. The purpose of this work is to propose a cumulation of deep learning techniques for the recognition of objects in video and to utilize an accessibility resource called audio-description to engender a narrative of detected information. The result is a video object narrative-based surveillance system that provides useful contextual information about the environment for blind and visually impaired people.