This paper provides a comprehensive review of deep learning-based self-supervised methods for general visual feature learning from images or videos. It begins by motivating the need for self-supervised learning due to the high cost and time-consuming nature of collecting and annotating large-scale datasets. The paper then outlines the general pipeline of self-supervised learning, which involves training ConvNets to solve predefined pretext tasks using pseudo labels generated from image or video attributes. The paper reviews common deep neural network architectures used in self-supervised learning, including AlexNet, VGG, ResNet, GoogLeNet, and DenseNet for image feature learning, and 2DConvNet-based, 3DConvNet-based, and LSTM-based methods for video feature learning. It also discusses the evaluation metrics and datasets commonly used in self-supervised learning, and provides a quantitative performance analysis of existing methods on benchmark datasets. Finally, the paper concludes with a discussion of future directions in self-supervised visual feature learning.This paper provides a comprehensive review of deep learning-based self-supervised methods for general visual feature learning from images or videos. It begins by motivating the need for self-supervised learning due to the high cost and time-consuming nature of collecting and annotating large-scale datasets. The paper then outlines the general pipeline of self-supervised learning, which involves training ConvNets to solve predefined pretext tasks using pseudo labels generated from image or video attributes. The paper reviews common deep neural network architectures used in self-supervised learning, including AlexNet, VGG, ResNet, GoogLeNet, and DenseNet for image feature learning, and 2DConvNet-based, 3DConvNet-based, and LSTM-based methods for video feature learning. It also discusses the evaluation metrics and datasets commonly used in self-supervised learning, and provides a quantitative performance analysis of existing methods on benchmark datasets. Finally, the paper concludes with a discussion of future directions in self-supervised visual feature learning.