EEG is a less costly and easy approach used to extract information regarding neuron functioning from the brain. There is a wide use of EEG to analyse various neurological disorder like dementia and Parkinson disease. A lot of researchers are working on detecting and predicting seizure using different images like MRI and FMRI images and using EEG signals by applying DL and ML methods. The proposed approach is focusing on designing algorithm for auto detection of seizure activity from the intracranial and surface EEG signals. The two different deep learning and machine learning model such as LSTM model and SVM classifier tools applied on the intracranial and surface EEG signals. The deep learning LSTM model attained 96.67% efficiency in auto-detection of seizure by analysing the EEG signal. The machine learning model was tested for auto-detection of seizures, and its utility with various kernels was compared. Both domain based event markers were evaluated from the EEG signal and utilized to train the SVM model for auto-seizure detection. The SVM model with the Shannon entropy event marker obtained 96.4% efficiency in auto-detection of nonseizure and seizure event from intracranial EEG signals. From surface EEG signal, the SVM model auto-detected seizures with highest 96.0% accuracy using time domain features.