Depression is considered among the most common mental disorders impacting the daily lives of people around the globe. Online social media has provided individuals the platforms to share their emotions and feelings; therefore, the depressive individuals may also be identified by processing the content. The advancements of natural language processing have provided the methods for depression detection from the content. This chapter intends to highlight the mainstream contributions for depression detection from the text contents shared on online social media. More precisely, hierarchical-based segregation is adopted for detailing the research contributions in the underlying domain. The top hierarchy depicts early detection and generic studies, followed by method, online social media, and community-based segregation. The subsequent hierarchy contains machine learning, deep learning, and hybrid studies in the context of method, Facebook, Twitter, and Reddit in terms of online social media, and general, literary, and geography as subhierarchies of community.