A seizure is an automatic change in behavior caused by a brief disturbance in the electrical activity of the brain. There are various sorts of seizures; some, known as focal seizures, only affect a specific area of the brain, while others, known as generalized seizures, impact the entire brain.Aberrant brain activity that causes seizures or episodes of unusual behavior, emotions, and sometimes an absence of consciousness is the characteristic of the neurological disorder epilepsy. A common sign of epilepsy, characterized by an imbalance in the brain's electrical rhythms, is recurrent seizures. If the activity before the seizure requires the highest level of concentration, such as driving a car, such seizures may cause situation-based accidents. As a result, severe circumstances like this cause fatal injuries.So detection and prediction of epileptic seizures has to be done as early as possible so that we can save a person's life from danger. Detection of seizures from EEG is a tedious task for neurologists manually because of complex and unexpected patterns and varying morphology of seizures. Large-scale EEG data can be used by ML and DL algorithms to effectively diagnose various seizure disorders and deliver results that are appropriate for neurologists. CNN-Convolutional Neural Network is used to classify the seizure and read accurate patterns from EEG. Utilize EEG inputs to identify seizures using CNN architecture, comparing time- and frequency-domain performance. In this paper, the author achieved an accuracy of 94.48% and 60.3% in frequency and time domain respectively for the CHB-MIT Scalp EEG database from physionet.org and 98.5% and 96.89% in frequency and time domain for the Preprocessed CHB-MIT Scalp EEG database from IEEE Data Portal submitted by the author for all 24 patients.