Extracting information from EEG signals in the human brain with the help of deep learning tools is a topic that is rapidly gaining popularity. Specifically, recognition of visual stimulus from brain Electroencephalography (EEG) signals has immense applications in the field of brain-computer interfacing. We propose a deep learning system for decoding visual stimuli from EEG signals. The proposed system comprises a classifier and a decoder. For the classifier module, we use a Deep Oscillatory Neural Network (DONN), which has hidden layers consisting of nonlinear neural oscillators. The features obtained from the last hidden layer of the classifier module are provided as input to the Decoder network which is a static feedforward network. The proposed system is trained on ThoughtViz EEG datasets. The proposed architecture exhibits superior classification performance compared to the performance reported in the literature.