Pedestrian Re-Recognition Algorithm Based on Optimization Deep Learning-Sequence Memory Model

Pedestrian re-recognition is an important research because it affects applications such as intelligent monitoring, content-based video retrieval, and human-computer interaction. It can help relay tracking and criminal suspect detection in large-scale video surveillance systems. Although the existing traditional pedestrian re-recognition methods have been widely applied to address practical problems, they have deficiencies such as low recognition accuracy, inefficient computation, and difficulty to adapt to specific applications. In recent years, the pedestrian re-recognition algorithms based on deep learning have been widely used in the pedestrian re-recognition field because of their strong adaptive ability and high recognition accuracy. The deep learning models provide a technical approach for pedestrian re-recognition tasks with their powerful learning ability. However, the pedestrian re-recognition method based on deep learning also has the following problems: First, the existing deep learning pedestrian re-recognition methods lack memory and prediction mechanisms, and the deep learning methods offer only limited improvement to pedestrian re-recognition accuracy. Second, they exhibit overfitting problems. Finally, initializing the existing LSTM parameters is problematic. In view of this, this paper introduces a revertive connection into the pedestrian re-recognition detector, making it more similar to the human cognitive process by converting a single image into an image sequence; then, the memory image sequence pattern reidentifies the pedestrian image. This approach endows deep learning-based pedestrian re-recognition algorithms with the ability to memorize image sequence patterns and allows them to reidentify pedestrians in images. At the same time, this paper proposes a selective dropout method for shallow learning. Selective dropout uses the classifier obtained through shallow learning to modify the probability that a node weight in the hidden layer is set to 0, thereby eliminating the overfitting phenomenon of the deep learning model. Therefore, this paper also proposes a greedy layer-by-layer pretraining algorithm for initializing LSTM and obtains better generalization performance. Based on the above explanation, this paper proposes a pedestrian re-recognition algorithm based on an optimized LSTM deep learning-sequence memory learning model. Experiments show that the pedestrian re-recognition method proposed in this paper not only has strong self-adaptive ability but also identifies the average accuracy. The proposed method also demonstrates a significant improvement compared with other mainstream methods because it can better memorize and learn the continuous motion of pedestrians and effectively avoid overfitting and parameter initialization in the deep learning model. This proposal provides a technical method and approach for adaptive pedestrian re-recognition algorithms.

Download Full-text

Arabic Handwriting Classification using Deep Transfer Learning Techniques

Pertanika Journal of Science and Technology ◽

10.47836/pjst.30.1.35 ◽

2022 ◽

Vol 30 (1) ◽

pp. 641-654

Author(s):

Ali Abd Almisreb ◽

Nooritawati Md Tahir ◽

Sherzod Turaev ◽

Mohammed A. Saleh ◽

Syed Abdul Mutalib Al Junid

Keyword(s):

Deep Learning ◽

Transfer Learning ◽

Learning Transfer ◽

Learning Model ◽

Text Recognition ◽

Suitable Model ◽

Learning Models ◽

Recognition Algorithms ◽

Learning Techniques ◽

Deep Learning Model

Arabic handwriting is slightly different from the handwriting of other languages; hence it is possible to distinguish the handwriting written by the native or non-native writer based on their handwriting. However, classifying Arabic handwriting is challenging using traditional text recognition algorithms. Thus, this study evaluated and validated the utilisation of deep transfer learning models to overcome such issues. Hence, seven types of deep learning transfer models, namely the AlexNet, GoogleNet, ResNet18, ResNet50, ResNet101, VGG16, and VGG19, were used to determine the most suitable model for classifying the handwritten images written by the native or non-native. Two datasets comprised of Arabic handwriting images were used to evaluate and validate the newly developed deep learning models used to classify each model’s output as either native or foreign (non-native) writers. The training and validation sets were conducted using both original and augmented datasets. Results showed that the highest accuracy is using the GoogleNet deep learning model for both normal and augmented datasets, with the highest accuracy attained as 93.2% using normal data and 95.5% using augmented data in classifying the native handwriting.

Download Full-text

Human Action Recognition Algorithm Based on Adaptive Initialization of Deep Learning Model Parameters and Support Vector Machine

IEEE Access ◽

10.1109/access.2018.2874022 ◽

2018 ◽

Vol 6 ◽

pp. 59405-59421 ◽

Cited By ~ 6

Author(s):

Feng-Ping An

Keyword(s):

Support Vector Machine ◽

Deep Learning ◽

Action Recognition ◽

Human Action Recognition ◽

Human Action ◽

Learning Model ◽

Recognition Algorithm ◽

Support Vector ◽

Model Parameters ◽

Deep Learning Model

Download Full-text

A Deep Learning Model to Recognize and Quantitatively Analyze Cold Seep Substrates and the Dominant Associated Species

Frontiers in Marine Science ◽

10.3389/fmars.2021.775433 ◽

2021 ◽

Vol 8 ◽

Author(s):

Haining Wang ◽

Xiaoxue Fu ◽

Chengqian Zhao ◽

Zhendong Luan ◽

Chaolun Li

Keyword(s):

Deep Learning ◽

Large Scale ◽

Recognition Accuracy ◽

Learning Model ◽

Cold Seep ◽

Cold Seeps ◽

Promising Tool ◽

And Function ◽

Associated Species ◽

Deep Learning Model

Characterizing habitats and species distribution is important to understand the structure and function of cold seep ecosystems. This paper develops a deep learning model for the fast and accurate recognition and classification of substrates and the dominant associated species in cold seeps. Considering the dense distribution of the dominant associated species and small objects caused by overlap in cold seeps, the feature pyramid network (FPN) embed into the faster region-convolutional neural network (R-CNN) was used to detect large-scale changes and small missing objects without increasing the number of calculations. We applied three classifiers (Faster R-CNN + FPN for mussel beds, lobster clusters and biological mixing, CNN for shell debris and exposed authigenic carbonates, and VGG16 for reduced sediments and muddy bottom) to improve the recognition accuracy of substrates. The model’s results were manually verified using images obtained in the Formosa cold seep during a 2016 cruise. The recognition accuracy of the two dominant species, e.g., Gigantidas platifrons and Munidopsidae could be 70.85 and 56.16%, respectively. Seven subcategories of substrates were also classified with a mean accuracy of 74.87%. The developed model is a promising tool for the fast and accurate characterization of substrates and epifauna in cold seeps, which is crucial for large-scale quantitative analyses.

Download Full-text

Deep Learning Model Selection of Suboptimal Complexity

Автоматика и телемеханика ◽

10.31857/s000523100001252-1 ◽

2018 ◽

pp. 129-147

Author(s):

Oleg Bakhteev ◽

◽

Vadim Strijov ◽

Keyword(s):

Deep Learning ◽

Model Selection ◽

Learning Model ◽

Deep Learning Model ◽

Selection Of

Download Full-text

Improving Sentiment Analysis using Hybrid Deep Learning Model

Recent Advances in Computer Science and Communications ◽

10.2174/2213275912666190328200012 ◽

2020 ◽

Vol 13 (4) ◽

pp. 627-640 ◽

Cited By ~ 1

Author(s):

Avinash Chandra Pandey ◽

Dharmveer Singh Rajpoot

Keyword(s):

Neural Network ◽

Deep Learning ◽

Sentiment Analysis ◽

Classification Accuracy ◽

Short Term Memory ◽

Computational Cost ◽

Extraction Process ◽

Learning Model ◽

Sentiment Classification ◽

Deep Learning Model

Background: Sentiment analysis is a contextual mining of text which determines viewpoint of users with respect to some sentimental topics commonly present at social networking websites. Twitter is one of the social sites where people express their opinion about any topic in the form of tweets. These tweets can be examined using various sentiment classification methods to find the opinion of users. Traditional sentiment analysis methods use manually extracted features for opinion classification. The manual feature extraction process is a complicated task since it requires predefined sentiment lexicons. On the other hand, deep learning methods automatically extract relevant features from data hence; they provide better performance and richer representation competency than the traditional methods. Objective: The main aim of this paper is to enhance the sentiment classification accuracy and to reduce the computational cost. Method: To achieve the objective, a hybrid deep learning model, based on convolution neural network and bi-directional long-short term memory neural network has been introduced. Results: The proposed sentiment classification method achieves the highest accuracy for the most of the datasets. Further, from the statistical analysis efficacy of the proposed method has been validated. Conclusion: Sentiment classification accuracy can be improved by creating veracious hybrid models. Moreover, performance can also be enhanced by tuning the hyper parameters of deep leaning models.

Download Full-text