Backbones-Review: Feature Extraction Networks for Deep Learning and Deep Reinforcement Learning Approaches

Abstract

To understand the real world using various types of data, Artificial Intelligence (AI) is the most used technique nowadays. While finding the pattern within the analyzed data represents the main task. This is performed by extracting representative features step, which is proceeded using the statistical algorithms or using some specific filters. However, the selection of useful features from large-scale data represented a crucial challenge. Now, with the development of convolution neural networks (CNNs), the feature extraction operation has become more automatic and easier. CNNs allow to work on large-scale size of data, as well as cover different scenarios for a specific task. For computer vision tasks, convolutional networks are used to extract features also for the other parts of a deep learning model. The selection of a suitable network for feature extraction or the other parts of a DL model is not random work. So, the implementation of such a model can be related to the target task as well as the computational complexity of it. Many networks have been proposed and become the famous networks used for any DL models in any AI task. These networks are exploited for feature extraction or at the beginning of any DL model which is named backbones. A backbone is a known network trained in many other tasks before and demonstrates its effectiveness. In this paper, an overview of the existing backbones, e.g. VGGs, ResNets, DenseNet, etc, is given with a detailed description. Also, a couple of computer vision tasks are discussed by providing a review of each task regarding the backbones used. In addition, a comparison in terms of performance is also provided, based on the backbone used for each task.

Vehicle Instance Segmentation Polygonal Dataset for Private Surveillance System

“…Naïve bayes [22], SVM [23], LBP [23], and KNN were common machine learning algorithms used for vehicle make and model classification. CNN architecture used for vehicle make and model classification involve transfer learning on prominent pretrained models like that of Alexnet, VGG, Resnet, and mobilenet [15]. Adding to this modified CNN networks were introduced such as residual squeezenet [24] which produced a higher rank-5 accuracy of 99.38.…”

Towards Robotic Marble Resin Application: Crack Detection on Marble Using Deep Learning

“…The selection of the most suitable network for feature extraction should not be completed by chance, since it is strongly related to the performance of the target task (e.g., segmentation) and it is responsible for, among other things, the computational complexity of the DL model. Many backbone networks have been developed and used in various DL models [49].…”

“…It should be noted that, in the literature, there are a lack of research works that aim to compare proposed feature extraction networks for their DL applications [49,50]. However, in computer vision tasks, the selection of the appropriate backbone network is critical; unsuitable backbones for specific applications can deteriorate the model's performance and be significantly complex and computationally costly.…”

Polygonal Dataset for Vehicle Instance Segmentation for Private Surveillance System

“…], SVM [21], LBP [21], and KNN were common machine learning algorithms used for vehicle make and model classification. CNN, used for vehicle make and model classification involve transfer learning on prominent pretrained models like that of Alexnet, VGG, Resnet, and mobilenet [15]. Adding to this modified CNN networks were introduced such as residual squeezenet [22] which produced a higher rank-5 accuracy of 99.38.…”