Research on Brain Visual Image Signal Recognition Method Based on Deep Neural Network

With the advancement of deep learning, the accuracy of image classification continues to rise. However, the generalization capability of deep learning methods, such as those applied in image classification, lags significantly behind that of the human brain. Therefore, in this study, we propose a novel approach driven by human brain activity for classifying visual image signals. Specifically, we utilize a non-invasive 64-channel EEG cap to capture the brain activity of six subjects as they view 40 images across five object categories. Subsequently, we employ two Long Short-Term Memory (LSTM) network structures to extract feature vectors from the EEG signals. Following this, we employ Convolutional Neural Network (CNN) to extract feature vectors from the images. Lastly, we conduct a regression task on the extracted EEG and image feature vectors, enabling the machine to perform visual image classification based on human brain-inspired features. Our LSTM-based approach achieves a maximum accuracy of 66.13% in distinguishing object categories, surpassing existing methods for learning EEG-based visual object representations. Regarding image classification, our human brain-driven approach demonstrates commendable performance, even comparable to robust CNN models. The dataset can be found on https://github.com/Goustjager0/CJY-dataset.