Stress is an individual’s psychological and physiological response to any circumstance or external environment perceived as a threat. In current times, stress is one of the leading causes of acute and chronic health issues in human beings. Advancements in smart devices and wearable technology in recent years have made it feasible to continuously monitor biomarkers linked with stress in a non-invasive manner. This paper aims to conduct a comprehensive study of various stress detection methods using deep learning techniques and explainable AI. The study focuses on the utilization of biomarkers measurable by smart devices like heart rate variability (HRV), blood oxygen levels, electrodermal activity (EDA), electrocardiogram (ECG), skin conductance, respiration rate, blood volume pressure (BVP), electromyogram (EMG), and body temperature. The paper also underlines existing datasets, ongoing challenges, and future research directions in the domain of stress detection.