The mental health detection has implemented using CNN, LSTM and BERT models. Among these 3 different models, Long Short Term Memory (LSTM) has acquire best accurecy of 97.78% where Convolutional Neural Network (CNN) has acquire 97.36% and Bidirectional Encoder Representations from Transformers (BERT) has acquire 82.51%.