A5001329945- Advanced Image and Video Retrieval Techniques
- Robotics and Sensor-Based Localization
- Visual Attention and Saliency Detection
- Energy Load and Power Forecasting
- Advanced Vision and Imaging
- Image Retrieval and Classification Techniques
- Video Surveillance and Tracking Methods
- 3D Shape Modeling and Analysis
- Computer Graphics and Visualization Techniques
- Dental Radiography and Imaging
- Linguistics and Cultural Studies
- Medical Imaging Techniques and Applications
- Stock Market Forecasting Methods
- Advanced Neural Network Applications
- Face Recognition and Perception
- Financial Markets and Investment Strategies
- Cancer-related molecular mechanisms research
- Seismic Imaging and Inversion Techniques
- Gaze Tracking and Assistive Technology
- Traditional Chinese Medicine Studies
Shandong University of Finance and Economics
2021-2024
Due to the interaction of many factors in stock market, price prediction has always been a challenging problem field machine learning. In particular, mutation market often have great impact on subsequent predictions. The existing models seldom consider impacts other stocks and points accuracy target stocks. Therefore, this paper presents new knowledge graph deep learning method combined with network focusing related points. First, features are obtained through ConvLSTM network. Second, is...
Although video deraining technology has achieved great success in recent years, extracting spatiotemporal feature representations across the domains of spatial and temporal successive frames, then performing modeling, restoring high-quality videos with rich details are still challenging tasks. In this paper, we use hybrid Transformer for first attempt rain removal tasks, propose a novel network based on transformer (VDN-HT) to aggregate global local accomplish deraining. extraction process,...
Abstract Automatic and accurate instance segmentation of teeth can provide important support for computer‐aided orthodontic work. Traditional methods tooth studies often ignore the rich structural features teeth. Capturing complete geometry as well morphological details a single remains challenge current studies. In this article, new deeplearning network based on capturing dependencies receptive field adjustment in cone beam computed tomography (CBCT) is proposed to achieve automatic dental...
Accurate segmentation of the tongue body is an important prerequisite for computer‐aided diagnosis. In general, size and shape are very different, color similar to surrounding tissue, edge fuzzy, some interfered by pathological details. The existing methods often not ideal image processing. To solve these problems, this paper proposes a symmetry edge‐constrained level set model combined with geometric features segmentation. Based on geometry tongue, novel initialization method proposed...
During the annotation procedure of salient object detection, researchers usually locate approximate location objects first and then process pixels that need to be finely annotated. Following this idea, we find existing methods have limited exploration for solving problem positioning objects. Furthermore, no effective solution has been proposed hard-sample related task. Therefore, propose dynamic scale-aware learning learn scale weights vary with different images solve problem. Second, design...
Tiny objects often have a small proportion of pixels in the image, leading to significant differences number positive and negative samples lack feature information. Accurately determining position category tiny remains huge challenge for object detection research. Therefore, we design an Adaptive Sample Assignment Strategy(ASAS) focusing enhancement module solve above two problems. Specifically, starting from study sample selection balance strategies objects, construct lightweight Object...
Efficient and accurate point cloud feature extraction is crucial for critical tasks such as 3D recognition semantic segmentation. However, existing global methods data often require designing different models input types (point clouds, voxels, maps). This article proposes an efficient plug-and-play non-iterative clustering method (NICM) to establish a unified paradigm suitable any type solve the above problems. The core idea of NICM construct connection between single other points based only...