Suicides are on the rise all across the world, and depression is a prevalent cause. As a result, effective diagnosis and therapy are required to lessen the symptoms of depression and anxiety. An electroencephalogram (EEG) is a device that measures and records electrical activity from the brain. It can be used to generate a precise assessment on the severity of depression and anxiety. Previous research has shown that EEG data and deep learning (DL) models can be used to diagnose various psychiatric disorders. As a result, this paper offers DeepNet, a DL-based convolutional neural network (CNN) for identifying EEG data from depressed, anxiety and healthy people. This study examines DeepNet's performance in two trials, namely the subject wise split and the record wise split. DeepNet's results have an accuracy of 0.9837, and when record wise split data is used, the area under the receiver operating characteristic curve (AUC) is 0.989.