Today's pupils frequently endure stress in their daily lives. They experience stress from a variety of sources, which is known to impair their performance. As a result of rising standards for academic performance, bad time management, and financial worries, stress has grown pervasive in the academic environment. Their physical and emotional health are both negatively impacted, which has a negative impact on their quality of life. If it goes unnoticed for a longer time, it increases the risk of sadness and suicidal thoughts. A approach that is non-invasive, exact, accurate, and trustworthy is required. The fact that electroencephalography (EEG) is a non-invasive process makes it the ideal tool. Moreover, it gets the feedback from the emotion released stress hormone, making it a gaining truth access tool to measure the stress. This project totally deals with the stress and the stress hormones are analysed and further the stress levels are detected and the students stress is detected. Fast Fourier Transform is being used to extract essential time-frequency characteristics from the EEG recordings after the EEG signal has been pre-processed to remove disturbances (FFT). Utilizing a split of the retrieved attributes, stress levels are determined using Deep Learning Convolution Neural Network (CNN) classifier. This technique is novel in that it modifies the CNN's convolution kernel to accommodate the input of EEG recordings. The classification accuracy of 86.4% is obtained. This system revealed that the efficiency to detect stress level using brain waves datasets .