Stress related disorders are extremely prevalent among workers in the corporate sector. The high demands and extended work hours of IT employment have led to an increase in stress among IT workers. In this paper a novel STress DEtection on it employees using YOLO deep learning (STEP-YOLO) has been proposed for detecting the Stress on IT Employees. Initially, the input images are pre-processed in this pre-processing image acquisition and image processing are done to enhance the image. The features are extracted from the pre-processed image it extracts the feature whether they are happy, sad or neutral. Finally, the objects are detected using the YOLOv5 technique to detect the stress on IT employees. The proposed STEP-YOLO achieves an accuracy rate of 99.35% and an overall accuracy rate of 0.94%, 1.61% and 0.7% which is comparatively higher than the existing methods such as UBFC-Phys, PSS and SAD. The proposed method takes 0.18 milliseconds to detect the stress on IT employees which is comparatively less than the existing methods respectively.