Networks are ubiquitous data structures in the real world. The accurate and efficient analysis of networks is critical to realizing many intelligent network-based services. However, most existing network analysis meth...
详细信息
ISBN:
(纸本)9781665426480
Networks are ubiquitous data structures in the real world. The accurate and efficient analysis of networks is critical to realizing many intelligent network-based services. However, most existing network analysis methods are developed for single networks and require a lot of labeled data, which is costly and time-consuming to acquire. Transfer learning has been widely accepted as an effective paradigm for tackling this problem by reusing the model trained on a supervised task. However, transfer learning on the non-euclidean network data has been investigated by no more than a few studies. To realize accurate node classification based on the knowledge learned from the labeled source network, this paper proposes to learn network-invariant and label-discriminative representations based on graph embedding and linear discriminant analysis. Specifically, we embed the source and target networks into adjacent vector spaces based on the graph attention network by minimizing the Sinkhorn distributional distances between their embeddings. To obtain label-discriminative features for learning better classification models, we then utilize a transferable linear discriminative analysis method to project the embeddings into label-discriminative subspaces. In the end, a support vector machine model trained on the labeled source network is utilized to classify the target nodes. Experiments on two pairs of networks illustrate that our method achieves the best performance and evaluates the effectiveness of the proposed modules.
Vehicle and driver detection in the highway scene has been a research hotspot in the field of object detection in recent years, and it is still a challenging problem in the research of traffic order and road safety. I...
Vehicle and driver detection in the highway scene has been a research hotspot in the field of object detection in recent years, and it is still a challenging problem in the research of traffic order and road safety. In this paper, we propose a novel end-to-end vehicle and driver detection method named VDDNet which is based on Cascade R-CNN and SENet. By introducing FPN structure and SENet attention mechanism in the backbone, the ability of the model to learn effective features is enhanced. It can improve the accuracy of detection in difficult scenes such as weak light, partial occlusion, and low picture resolution. The test results based on the database of highway traffic vehicle and drivers constructed by the Jiangsu Provincial Public Security Department. It shows that the detection method has the AP rate of 91.3% and the Recall rate of 92.4%, which demonstrates the effectiveness of the proposed method in complex highway environments.
In this paper, we present a simple image depth level estimation algorithm. From the dark channel prior theory, an estimate of the air transmittance in the image is calculated. In wild surveillance, the disparity in th...
In this paper, we present a simple image depth level estimation algorithm. From the dark channel prior theory, an estimate of the air transmittance in the image is calculated. In wild surveillance, the disparity in the image poses a huge challenge for smoke detection and other video analysis tasks. Appropriate depth level estimation provide significant prior knowledge for subsequent identification and detection. For landscape images, we can approximate the air transmittance to depth information for histogram analysis. The depth value is segmented by a multi-threshold segmentation algorithm, and the resulting image can be used for forest fireworks detection and the like. This method does not rely on samples and classifiers, and the algorithm does not require training. The final experimental results show that the depth level estimation of a single landscape image based on the dark channel prior can achieve good results.
Chiral metamaterials are manmade structures with extraordinary mechanical properties derived from their special geometrical design instead of the chemical composition. To make the mechanical deformation programmable, ...
详细信息
Face detection and location technique is a hot research direction during recent years. Especially, driver face detection on highway is still a challenging problem in social safty deserving research. This paper propose...
Face detection and location technique is a hot research direction during recent years. Especially, driver face detection on highway is still a challenging problem in social safty deserving research. This paper proposes a novel algorithm based on the improved Multi-task Cascaded Convolutional Networks (MTCNN) and Support Vector Machine (SVM) to realize accurate face region detection and feature location of driver's face on highway, predicting face and feature location via a coarse-to-fine pattern. The proposed algorithm is verified under various complex highway conditions. Experimental results show that the proposed model shows satisfied performance compared to other state-of-the-art techniques used in driver face detection and alignment, keeping robust to the occlusions, varying pose and extreme illumination on highway.
This paper focuses on smoke detection in forest environments. In this paper, dark channel prior and OTSU based multi-threshold are used to find the disturbances such as sky and haze. These regions are blocked to reduc...
This paper focuses on smoke detection in forest environments. In this paper, dark channel prior and OTSU based multi-threshold are used to find the disturbances such as sky and haze. These regions are blocked to reduce false alarm rate. A motion detection method based on frame difference is adapted to find the motion objects. Color moments, HOG and LBP are chosen as features of smoke and SVM is used as the classification. To reduce more false alarms, the motion regions are classified for several consecutive frames. They won't be regarded as smoke regions unless M frames of them are classified as smoke ones. Experiment results showed that the proposed method can detect smoke in video effectively and work real-timely.
This paper aims to propose a modelling framework for service quality evaluation of passenger hub under fuzzy environment, i.e., when valuators cannot quantify all criteria accurately. The research problem is to evalua...
详细信息
This paper aims to propose a modelling framework for service quality evaluation of passenger hub under fuzzy environment, i.e., when valuators cannot quantify all criteria accurately. The research problem is to evaluate the operation status of passenger hub that maximizes the service level, including not only quantitative index but also qualitative index. A mixed-model combined the Grey Relational Analysis (GRA) and VIseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR) has been developed to solve the problem in a fuzzy environment where both criteria and weights could be fuzzy sets. An empirical study for evaluating service quality of three passenger hubs in different times was put forth to illustrate an application of the proposed model. The results reveal a number of interesting insights into service level evaluation, namely, this study has instructive significance in management practice.
Chiral metamaterials are manmade structures with extraordinary mechanical properties derived from their special geometrical design instead of the chemical composition. To make the mechanical deformation programmable, ...
详细信息
Face recognition technology is widely applied in daily life, but in most methods, similarity or affine transformation is employed to align face images according to five facial landmarks. The face alignment module is i...
Face recognition technology is widely applied in daily life, but in most methods, similarity or affine transformation is employed to align face images according to five facial landmarks. The face alignment module is implemented independently, thus it's difficult in end-to-end training. In this paper, the main purpose is to design a towards end-to-end trainable face recognition method based on indoor scenes. Due to that spatial transformer can implement any parametrizable transformation, we joint it with recognition network, making end-to-end training possible. Simultaneously, any prior knowledge on facial landmarks isn't required. The model jointly with spatial transformer can achieve 0.3% higher accuracy than similarity transformation. Most downsampling methods ignore the sampling theorem, making convolutional networks not shift-invariant. We replace max-pooling by MaxBlurPool in spatial transformer network, and the accuracy is improved by 0.25%.
Deep learning-based no-reference image quality assessment (NR-IQA) algorithm requires massive training data and labels, but data scarcity usually exists in practical applications. Most of the previous metrics used pre...
详细信息
ISBN:
(纸本)9781665426480
Deep learning-based no-reference image quality assessment (NR-IQA) algorithm requires massive training data and labels, but data scarcity usually exists in practical applications. Most of the previous metrics used pre-trained networks to solve this problem by fine-tuning on the target tasks. Unfortunately, the network trained on other tasks cannot be directly applied to IQA tasks and failed when encountering complex distortions. To reduce dependence on labeled data and improve generalization ability of unknown distortion types, a NR-IQA metric based on Meta-SGD is proposed, which can learn the quick adaptability of humans and obtain general meta-knowledge for distortion evaluation. Specifically, we collected a great many NR-IQA tasks with different distortion types to pre-train the model; then a meta-learning metric based on optimization, Meta-SGD, is proposed to acquire the meta-knowledge when evaluating images quality with various known distortions; finally, the obtained pre-trained meta-model can be directly applied to the new NR-IQA tasks only by fine-tuning a few images. Experiments on the TID2013 database demonstrate that the model after meta-learning obviously outperforms than some existing methods, and the average SRCC value for all types of distortion performance tests can reach 0.91.
暂无评论