Depression is a widespread mental health disorder and a leading cause of suicide today. According to reports, millions of people are affected by depression and suicide every year. The World Health Organization (WHO) 2025 survey report stated that more than 332 million people are suffering from depression. Suicidal cases related to depression can be prevented by early diagnosis of the disease. Structured clinical interviews, self-report questionnaires, and behavioral observation are traditional methods for detecting depression, but these methods have limitations, such as being time-consuming and lacking validation. Therefore, this study aims to present a depression detection method based on Multilayer Perceptron (MLP), a deep learning approach. The PHQ-9 dataset was used for experimentation to detect the depression at early stage. To test the performance of the proposed method, we compared it to the six commonly used machine learning models viz. Random Forest (RF), Support Vector Machine (SVM), Linear Regression (LR), CatBoost, and Convolutional Neural Network (CNN). Several experiments were conducted by tuning hyperparameters such as in RF parameter setting max_depth = 30 and n_estimator limited to 300, regularization parameter in SVM is set to minimum to enforce stronger regular regularization and GridSearchCV technique is used to identify the best combination of parameters and validate results of proposed method, we used metrics accuracy and execution time. The comparison shows that the proposed method based on MLP model outperformed over other commonly used ML based methods.