Small sample image recognition using improved Convolutional Neural Network

Zhang, Jiajia; Shao, Kun; Luo, Xing

doi:10.1016/j.jvcir.2018.07.011

Cited by 61 publications

(16 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The next most important step is to choose the classifier that accepts the feature vector from CNN feature learning module and generates output labels. In [46][47][48], the authors claimed that the SVM is a strong and fast classifier for real-time classification applications and great attention has been paid to the fusion of neural networks and SVM [49,50]. That is why the same is used in our proposed system.…”

Section: License Plate Character Recognitionmentioning

confidence: 99%

An Adaptive Approach for Multi-National Vehicle License Plate Recognition Using Multi-Level Deep Features and Foreground Polarity Detection Model

Asif

Khan

2020

Applied Sciences

View full text Add to dashboard Cite

License plate recognition system (LPR) plays a vital role in intelligent transport systems to build up smart environments. Numerous country specific methods have been proposed successfully for an LPR system, but there is a need to find a generalized solution that is independent of license plate layout. The proposed architecture is comprised of two important LPR stages: (i) License plate character segmentation (LPCS) and (ii) License plate character recognition (LPCR). A foreground polarity detection model is proposed by using a Red-Green-Blue (RGB) channel-based color map in order to segment and recognize the LP characters effectively at both LPCS and LPCR stages respectively. Further, a multi-channel CNN framework with layer aggregation module is proposed to extract deep features, and support vector machine is used to produce target labels. Multi-channel processing with merged features from different-level convolutional layers makes output feature map more expressive. Experimental results show that the proposed method is capable of achieving high recognition rate for multinational vehicles license plates under various illumination conditions.

show abstract

Section: License Plate Character Recognitionmentioning

confidence: 99%

An Adaptive Approach for Multi-National Vehicle License Plate Recognition Using Multi-Level Deep Features and Foreground Polarity Detection Model

Asif

Khan

2020

Applied Sciences

View full text Add to dashboard Cite

show abstract

“…CNN known with its tremendous performance in image processing in and recognition [18]. As it nature as a part of multi-layer perceptron classifier, training stage is use to construct the weights and biases of the involved convolution kernels.…”

Section: E Convolutional Neural Network (Cnn) Classificationmentioning

confidence: 99%

CNN and SVM Based Classifier Comparation to Detect Lung Nodule In Computed Tomography Images

Sentana¹,

Asri²,

Jawas³

et al. 2018

Proceedings of the International Conference on Science and Technology (ICST 2018)

View full text Add to dashboard Cite

Convolutional Neural Networks (CNN) are natural based classification algorithm that combine Multiple Layer Perceptron (MLPs). Meanwhile, support vector machines (SVM) is a mathematical-based classification algorithm that naturally have supervised learning models. In some research related to image processing, each algorithm has its owned supremacy as well as the drawback. None of the previous studies compare both algorithm when they are utilized to detect nodule located in the pulmonary or lung images produced by Computed Tomography (CT) scan. Hence, this research comparing the two algorithms in case of lung nodule detection in CT images, since detecting lung nodule in CT images is still challenging. SVM-based classifier is preceded by feature extraction as its common behavior of mathematical based classifier. There are three algorithms use to conduct feature extraction process, namely Hu moment invariant, Haralick and Color Histogram extraction. In the opposite, CNN-based classifier consists of three layers convolution for training and testing steps. The result shows that SVM has better results than CNN in case of computing speed. Meanwhile have a better accuracy in detecting lung nodule. The results of the test analysis show that the extractor feature when preprocessing conduct before being classified by SVM makes the computing process faster. The accuracy of SVM-based classifier can be improved by adjusting some computation variables in feature extraction stages, such as adding more bins in the color histogram extraction. Those adjustment will lead to more computation times.

show abstract

“…Based on the VGG-Net model, combined with the traditional optical flow characteristics, the optical flow graph is also regarded as an image, and a dual data stream deep convolutional neural network is proposed [16,17]. By using the trajectory features in the traditional method, by tracking the trajectory of the optical flow, the convolutional features are concentrated in the areas where the motion is more significant, and then the extracted features are subjected to a one-step down-sampling operation to obtain the final convolutional feature map, Using linear SVM as a classifier for action recognition [18,19]. It is proposed to use Recurrent Neural Network (RNN) to establish a time series model for video sequences.…”

Section: Introductionmentioning

confidence: 99%

Human Motion Gesture Recognition Algorithm in Video Based on Convolutional Neural Features of Training Images

2020

IEEE Access

View full text Add to dashboard Cite

The main work of human motion gesture recognition is to recognize and analyze the behavior of human objects in the video. Although the current research in the field of human motion gesture recognition has achieved certain results, the human motion gesture recognition in real life scenes has great effects due to factors such as camera movement, target scale transformation, dynamic background, viewing angle, and illumination. This article first proposes a new method of constructing human motion posture features to describe human behavior. This method is based on deep convolutional neural network features and topic models. Experiments have verified that compared with the traditional feature map extracted from the convolutional neural network fully connected layer, the feature map extracted from the convolutional neural network convolutional layer is not only lower in dimension but also has higher discrimination. Secondly, based on the feature map of the convolutional neural network, the training map downsampling strategy is used to overcome the interference caused by the object's scale change and shape change. Finally, based on the basketball gesture recognition method, the behavior performance of the legs and arms in 9 basketball actions of walking, running, jumping, standing dribbling, walking dribbling, running dribbling, shooting, passing and receiving is analyzed. As well as the corresponding signal waveform characteristics, a two-stage data division method for basketball is proposed. The unit action data is extracted for analysis to realize feature extraction. In order to select the most suitable classifier for basketball gesture recognition, the constructed feature vector uses four Different classifiers are trained to construct different classifiers to realize the division of actions. INDEX TERMS Convolutional neural networks; training Images; human motion gesture recognition; classifier; feature vector Xiangui Bu was born in WeiShan, China, in 1981. Master of Physical Education, master of Education doctor, associate professor, master's supervisor, head of boxing course of Shandong Sport University, the first famous teacher of Shan dong Sport University. Visiting scholar of

show abstract

Small sample image recognition using improved Convolutional Neural Network

Cited by 61 publications

References 23 publications

An Adaptive Approach for Multi-National Vehicle License Plate Recognition Using Multi-Level Deep Features and Foreground Polarity Detection Model

An Adaptive Approach for Multi-National Vehicle License Plate Recognition Using Multi-Level Deep Features and Foreground Polarity Detection Model

CNN and SVM Based Classifier Comparation to Detect Lung Nodule In Computed Tomography Images

Human Motion Gesture Recognition Algorithm in Video Based on Convolutional Neural Features of Training Images

Contact Info

Product

Resources

About