Depression is the most common mental illness of human due to multiple reasons and often lead the sufferers to commit suicide. As a consequence, it is necessary to detect the depression state of the sufferer. Depression detection aims at predicting the depression level of the sufferer via information collected from social platforms. The information expressed by the users in social media is in the form of text, image, video, and audio. Gathering the textual information and visual information helps make accurate predictions and inferences about depression. Extracting the relevant information from the different forms is a challenging task in depression detection traditional methods.
   Multimodal extraction is the approach to extract the appropriate features from the different data modalities of information. Multimodal extraction combines the different forms of information to accurately extract the features that help in predicting the depression state of the sufferer. Integrating multimodal extraction for depression detection improves the robustness, performance, accuracy of the model.