Electroencephalograms (EEGs) are neuro-electrophysiology signals that are frequently employed as a diagnostic instrument for determining the brain activity that comes with seizures. A patient’s diagnosis can be improved with the use of precise epilepsy detection and classification. This research proposes an automated EEG classification approach using one-dimensional convolutional recurrent neural network (1D-CRNN) with attention mechanism (AM), which is a combination of convolutional neural network (CNN) and recurrent neural network (RNN). To overcome the vanishing gradient issues, 1D-CRNN is integrated with a dot-product AM. The AM helps the network to capture more epileptic regions and neglect the non-epileptic regions for accurate analysis. The proposed methodology uses a University of California (UCI) dataset for recorded EEG signals. The signals are sampled initially by preprocessing the samples using normalization and one-hot encoding. The 1D-CRNN has convolutional layers which are used for extracting spatial features and for capturing long-term dependencies. When applying the AM, the CRNN gives crucial weight features, thereby enhancing the method’s accuracy. The experimental results show that the proposed 1D-CRNN-AM approach has achieved better results with 99.24% of accuracy, 97.63% of precision, 97.23% of recall, and 98.45% of F1-score when compared with existing methods RNN-bidirectional long short-term memory (RNN-BiLSTM) and deep canonical sparse auto-encoder-based epileptic seizure detection and classification (DCSAE-ESDC) on evaluating with UCI dataset.