In order to solve the problem of huge labeling cost for person re-identification, a method of one-shot video-based person re-identification with multi-loss learning and joint metric was proposed. Aiming at the problem that the number of label samples is small and the model obtained is not robust enough, a Multi-Loss Learning (MLL) strategy was proposed. In each training process, different loss functions were used for different data to optimize and improve the discriminative ability of the model. Secondly, a Joint Distance Metric (JDM) was proposed for label estimation, which combined the sample distance and the nearest neighbor distance to further improve the accuracy of pseudo label prediction. JDM solved the problems of the low accuracy of label estimation for unlabeled data, and the instability in the training process caused by the unlabeled data not fully utilized. Experimental results show that compared with the one-shot progressive learning method PL (Progressive Learning), the rank-1 accuracy reaches 65.5% and 76.2% on MARS and DukeMTMC-VideoReID datasets when the ratio of pseudo label samples added per iteration is 0.10, with the improvement of the proposed method of 7.6 and 5.2 percentage points, respectively.
Current research on Person Re-Identification (Re-ID) mainly concentrates on short-term situations with person’s clothing usually unchanged. However, more common practical cases are long-term situations, in which a person has higher possibility to change his clothes, which should be considered by Re-ID models. Therefore, a method of person re-identification with cloth changing based on joint loss capsule network was proposed. The proposed method was based on ReIDCaps, a capsule network for cloth-changing person re-identification. In the method, vector-neuron capsules that contain more information than traditional scalar neurons were used. The length of the vector-neuron capsule was used to represent the identity information of the person, and the direction of the capsule was used to represent the clothing information of the person. Soft Embedding Attention (SEA) was used to avoid the model over-fitting. Feature Sparse Representation (FSR) mechanism was adopted to extract discriminative features. The joint loss of label smoothing regularization cross-entropy loss and Circle Loss was added to improve the generalization ability and robustness of the model. Experimental results on three datasets including Celeb-reID, Celeb-reID-light and NKUP prove that the proposed method has certain advantages compared with the existing person re-identification methods.
Session-based recommendation aims to predict the user’s next click behavior based on the click sequence information of the current user’s anonymous session. Most of the existing methods realize recommendations by modeling the item information of the user’s session click sequence and learning the vector representation of the items. As a kind of coarse-grained information, the item category information can aggregate the items and can be used as an important supplement to the item information. Based on this, a Session-based Recommendation model of Multi-granular Graph Neural Network (SRMGNN) was proposed. Firstly, the embedded vector representations of items and item categories in the session sequence were obtained by using the Graph Neural Network (GNN), and the attention information of users was captured by using the attention network. Then, the items and item category information given by different weight values of attention were fused and input into the Gated Recurrent Unit (GRU). Finally, through GRU, the item time sequence information of the session sequence was learned, and the recommendation list was given. Experiments performed on the public Yoochoose dataset and Diginetica dataset verify the advantages of the proposed model with the addition of item category information, and show that the model has better effect compared with all the eight models such as Short-Term Attention/Memory Priority (STAMP), Neural Attentive session-based RecomMendation (NARM), GRU4REC on the evaluation indices Precision@20 and Mean Reciprocal Rank (MRR)@20.
In order to solve the problems that the efficiency is low when the segment of three-dimensional Computed Tomography Angiography (CTA) coronary arteries images with complex structure and small region of interest, a segmentation algorithm combining region growing and graph cut was proposed. Firstly, a method of region growing based on threshold was used to divide images into several regions, which removed irrelevant pixels and simplified structure and protruded regions of interest. Afterwards, according to grey and space information, simplified images were constructed as a network diagram. Finally, network diagram was segmented with theory of graph cut, so the segmentation image of coronary arteries was got. The experimental results show that, compared with traditional graph cut, the increment for the segmentation efficiency is about 51.7%, which reduces the computational complexity. On the aspect of rendering quality, target areas for segmentation images of coronary arteries is complete, which is helpful for doctors to analyze the lesion correctly.
In order to filter out Gaussian noise and impulse noise at the same time, and get high resolution image in super-resolution reconstruction, a method with L1 and L2 mixed norm and Bilateral Total Variation (BTV) regularization was proposed for sequence images super-resolution. Firstly, multi-resolution optical flow model was used to register low-resolution sequence images and the registration precision was up to sub-pixel level, then the complementary information was used to raise image resolution. Secondly, taking advantage of L1 and L2 mixed norm, BTV regularization algorithm was used to solve the ill-posed problem. Lastly, the proposed algorithm was used to sequence images super-resolution. Experimental results show that the method can decrease the mean square error and increase Peak Signal-to-Noise Ratio (PSNR) by 1.2 dB to 5.2 dB. The algorithm can smooth Gaussian and impulse noise, protect image edge information and improve image identifiability, which provides good technique basis for license plate recognition, face recognition, video surveillance, etc.
To solve the problems of three-dimensional clipping and Multi-Planar Reformation (MPR) that only the geometrical information of the tissues or organs can be obtained and the structure of a curving organ cannot be displayed in a single image, a Curved Planar Reformation (CPR) algorithm based on MPR to extract the outline was proposed to reform the coronary artery. Firstly, the discrete points expressing the outline of the coronary artery were extracted by using MPR. Afterwards, the Cardinal interpolation was used to get smooth outline fitting curve. Secondly, the outline was projected along the interested direction to get the scanning curved planar. Finally, the scanning curved planar corresponding to the volume data of the cardiac was displayed, so the CPR image of artery was got. The experimental results show that, compared with three-dimensional clipping method and three-dimensional data field method, the increment for the extracting speed of the coronary artery outline is about 4 to 6 frames per second, and the rendering time is shorter. On the aspect of rendering quality, compared with three-dimensional segmentation method, the image of coronary artery curved plane is clear and complete, which is helpful for doctors to analyze the lesion clearly and satisfies the demands of actual clinical diagnosis.
Concerning the deficiency in scalability of the traditional hierarchical clustering algorithm when dealing with large-scale text, a parallel hierarchical clustering algorithm based on the MapReduce programming model was proposed. The vertical data partitioning algorithm based on the statistical characteristic of the components group of text vector was developed for data partitioning in MapReduce. Additionally, the sorting characteristics of the MapReduce were applied to select the merge points, making the algorithm be more efficient and conducive to improve clustering accuracy. The experimental results show that the proposed algorithm is effective and has good scalability.