Spatio-Temporal Transformer for Online Video UnderstandingDu, Zexu; Zhang, Guoliang; Lu, Weijiang; Zhao, Ting; Wu, Peng
doi: 10.1088/1742-6596/2171/1/012020pmid: N/A
Leading methods in the field of online video understanding try to extract useful information from the spatial and temporal dimensions of an input video. But they are suffering from two problems: (1) These methods can only extract local video information, and cannot relate to the important features of the temporal context in the video. (2) Although some methods can quickly process the information of each frame in the video, the processing efficiency of the whole video is not good, so this type of method cannot be applied to online video understanding tasks. This article introduces a Transformer-based network, which considers spatial and temporal content, and can quickly process each video at the same time. Our approach can efficiently handle up to 170 videos with hundreds of frames per second for action classification. Our method achieve 10 to 90 times faster than existing methods on the action classification datasets.
A Q-matrix Model Based on Binarized Neural NetworkLi, Jingjiang; Jiang, Chong; Ye, Shiwei
doi: 10.1088/1742-6596/2171/1/012037pmid: N/A
Q-matrix theory plays an important role in the field of cognitive diagnosis assessment. It is time-consuming and laborious for experts to define Q-matrix from a large number of data. In order to deal with this problem, this article comes up with a Q- matrix generation model based on binarized neural network. We combine the neural network with the Boolean operation relations of Q-matrix, A matrix and R matrix in item response theory, so that the model can mine Q-matrix more effectively.
Hybrid Pruning Method Based on Convolutional Neural Network Sensitivity and Statistical ThresholdGuo, Changyi; Li, Ping
doi: 10.1088/1742-6596/2171/1/012055pmid: N/A
The hybrid pruning algorithm can not only ensure the precision of the network, but also achieve a good balance between pruning ratio and computation. However, traditional pruning algorithms use coarse-grained or fine-grained pruning networks, which have the tradeoff problem between pruning rate and computation amount. To this end, this paper presents. A hybrid pruning method of sensitivity and statistical threshold. Firstly, coarse-grained pruning is carried out on the network, and a fast sensitivity test is conducted on the convolutional layer of the network to determine the channels that need pruning within the tolerance range of network precision decline. Then, fine-grained pruning is performed on the network. Count the weights of the pruned network, calculate the thresholds of the weights of each layer, and delete the weights less than the thresholds so as to further reduce the size of the network and the amount of calculation. Hybrid pruning performs very well in AlexNet and Resnet networks. In particular, the method proposed in this paper is used in CIFAR-10 dataset, and the compression of FLOPs is 60%, while the compression of parameter number is nearly 80%. Compared with single pruning method, Hybrid pruning is better.
Semi-supervised LDA and Multi-distance Metric Learning for Person Re-identificationLi, Bin; Ding, Haoyang; Zhou, Mengjing
doi: 10.1088/1742-6596/2171/1/012054pmid: N/A
The problem of person re-identification has attracted a lot of attention in the field of machine vision. In practice, the non-overlapping sample images change drastically and the sample size is small, which makes the metric model overfitting phenomenon. In this paper, based on the k-NN and the sample normality property, we propose a resampling linear discriminant analysis (LDA) algorithm to suppress the local constraints caused by small samples, then train it to obtain the person re-identification metric learning model. A semi-supervised LDA algorithm with semi-supervised characteristics is developed by optimizing the inter-class scatter for weighting. A joint distance metric-based approach is also proposed to learn both the Mahalanobis distance and Euclidean distance. The improved algorithm is tested on the VIPeR and CUHK01 datasets, and the results indicate that, despite the change in the total number of training samples, the algorithm in this paper shows high recognition accuracy.
Finger Vein Image ROI Extraction Based on Active Contour MethodWang, Yifan; Lu, Huimin; Gao, Ruoran; Wang, Guizeng
doi: 10.1088/1742-6596/2171/1/012070pmid: N/A
Under the background of the novel coronavirus pneumonia outbreak in the world, unrestricted and contactless finger vein collection devices have significantly improved public health safety. However, due to the unfixed position of the finger and the open or semi-open characteristics of the acquisition device, it is inevitable to introduce plenty of factors that affect the recognition performance, such as low contrast, uneven illumination and edge disappearance. In view of these practical problems, we propose a method for ROI extraction of finger vein images that combines active contour method and morphological post-processing operations. This method starts from the local segmentation, and finally completes the acquisition of finger masks at the global level, and then combines some morphological operations to achieve precise extraction of finger masks. We designed and conducted plenty of comparison experiments on the proposed algorithm and the current mainstream finger vein image ROI extraction methods on three public available finger vein datasets. Experimental results show that our method accurately extracts the complete finger region mask and achieves the best matching accuracy on all datasets.
Research on Safety Helmet Detection Algorithm of Power Workers Based on Improved YOLOv5Fu, Desu; Gao, Lin; Hu, Tao; Wang, Shukun; Liu, Wei
doi: 10.1088/1742-6596/2171/1/012006pmid: N/A
The traditional helmet detection algorithm in power industry has low precision and poor robustness. In response to this problem, the helmet detection algorithm based on improved YOLOv5 (You only look once) is put forward in this paper. Firstly, the YOLOv5 network structure is improved. By increasing the size of the feature map, one scale is added to the original three scales, and the added 160*160 feature map can be used for the detection of small targets; Secondly, the K-means is used for re-clustering the helmet data set to get more suitable priori anchor boxes. The experimental results illustrate that the average accuracy of the improved YOLOv5 algorithm is increased by 2.9% and reaching 95% compared with the initial model, and the accuracy of helmet recognition is increased by 2.4% and reaching 94.6%. This algorithm reduces the rates of missing detection and misdetection of small target detection in original network, and has strong practicability and advanced nature. It can satisfy the requirements of real-time detection and has a certain role in promoting the safety of power industry.
Research on Visual Odometer of Wheeled Robot with Motion ConstraintsChenggong, Wang; Gen, Li; Xi, Wang; Youyi, Zhang
doi: 10.1088/1742-6596/2171/1/012064pmid: N/A
In this paper, aiming at the planar motion characteristics of wheeled robots, the Red Green Blue Depth Camera (RGBD camera) was used as an image acquisition device to reduce the motion of wheeled robots to two-dimensional plane processing, which simplified the calculation method of visual mileage of planar wheeled robots. In addition, according to the two-dimensional motion characteristics of the robot camera, the contour constraint condition of feature point matching between frames was proposed, and the linear constraint condition was proposed by linearizing the motion equation of the camera between frames. The research showed that the contour constraint condition and linear constraint condition could be used to screen the mismatched feature points, this condition could not only screen the mismatched point pairs of color images, but also screen the matching point pairs with correct color image matching but large depth error, which provided high quality matching point pairs for inter-frame motion estimation of visual odometer. Finally, the filtered matching point pairs were reduced in dimension and combined with the two-dimensional Iterative Closest Point (ICP) algorithm to estimate the trajectory of the robot camera. The experimentation results showed that compared with the original three-dimensional ICP algorithm, the combination of contour constraint condition, linear constraint condition and two-dimensional ICP algorithm could significantly improve the computational speed of visual odometry.
A Survey of Entity Alignment of Knowledge Graph Based on Embedded RepresentationHuang, Jing; Wang, Jiaqi; Li, Yahui; Zhao, Wenbin
doi: 10.1088/1742-6596/2171/1/012050pmid: N/A
This paper summarizes the main methods of knowledge representation learning. Representation learning represents the entity information of the knowledge graph as a low dimensional vector. Its vector dimension is low, which helps to improve the computational efficiency and make full use of the semantic information between entities. In order to embed two KGs into a unified semantic space, joint embedding is used to achieve this goal. With the development of research, there are many improved embedding methods, such as iteration, multi view embedding, knowledge graph entity alignment based on graph neural network and so on.
Residual Network and Embedding Usage: New Tricks of Node Classification with Graph Convolutional NetworksChi, Huixuan; Wang, Yuying; Hao, Qinfen; Xia, Hong
doi: 10.1088/1742-6596/2171/1/012011pmid: N/A
Graph Convolutional Networks (GCNs) and subsequent variants have been proposed to solve tasks on graphs, especially node classification tasks. In the literature, however, most tricks or techniques are either briefly mentioned as implementation details or only visible in source code. In this paper, we first summarize some existing effective tricks used in GCNs mini-batch training. Based on this, two novel tricks named GCN_res Framework and Embedding Usage are proposed by leveraging residual network and pre-trained embedding to improve baseline’s test accuracy in different datasets. Experiments on Open Graph Benchmark (OGB) show that, by combining these techniques, the test accuracy of various GCNs increases by 1.21%∼2.84%. We open source our implementation at https://github.com/ytchx1999/PyG-OGB-Tricks.
Automatic Information Extraction for Financial Events by Integrating BiGRU and Attention MechanismLu, Jiaheng; Liu, Weirong
doi: 10.1088/1742-6596/2171/1/012001pmid: N/A
In this paper, an information extraction method for financial events written in Chinese is proposed. The core entities of the causes and results, as well as the verbs and conditions are extracted from the financial events reported by websites. The method takes the original words and the part-of-speech of words as two inputs. BERT encoder is utilized to transform the original sentences to word-embedding vectors, which then are send to BiGRU to extract the sematic features. And a full-connected network is overlapped on BiGRU to reduce the impact of “covariate shift”. For the second input, the original sentences are cut by Chinese cut-word tool ‘jieba’ to get the part-of-speech of words, which are then transformed by self-attention mechanism to get global dependencies. The two outputs for word-embedding vectors and part-of-speech of words are combined and then decoded by CRF. Finally, the Viterbi algorithm is utilized to get the best sequences. The experiment results validate the effectiveness of the proposed method.