A5021302609- AI in cancer detection
- Video Coding and Compression Technologies
- Advanced Vision and Imaging
- Domain Adaptation and Few-Shot Learning
- Radiomics and Machine Learning in Medical Imaging
- Advanced Neural Network Applications
- Advanced Data Compression Techniques
- Advanced Image Processing Techniques
- Transportation Planning and Optimization
- Multimodal Machine Learning Applications
- Digital Imaging for Blood Diseases
- Advanced Image and Video Retrieval Techniques
- Image Retrieval and Classification Techniques
- Medical Imaging and Analysis
- Generative Adversarial Networks and Image Synthesis
- Image and Video Quality Assessment
- Adversarial Robustness in Machine Learning
- COVID-19 diagnosis using AI
- Machine Learning and Data Classification
- Brain Tumor Detection and Classification
- Acoustic Wave Phenomena Research
- Traffic control and management
- Image Enhancement Techniques
- Traffic Prediction and Management Techniques
- Colorectal Cancer Screening and Detection
Beijing University of Posts and Telecommunications
2018-2025
Beijing Jiaotong University
2023-2025
Chongqing University of Posts and Telecommunications
2025
Donghua University
2025
Hefei University of Technology
2023-2024
Qinghai New Energy (China)
2022-2024
Central China Normal University
2023-2024
Qinghai University
2022-2024
Fudan University
2023
Beijing Chao-Yang Hospital
2023
It is very challenging for various visual tasks such as image fusion, pedestrian detection and image-to-image translation in low light conditions due to the loss of effective target areas. In this case, infrared visible images can be used together provide both rich detail information paper, we present LLVIP, a visible-infrared paired dataset low-light vision. This contains 33672 images, or 16836 pairs, most which were taken at dark scenes, all are strictly aligned time space. Pedestrians...
Abstract Background Breast cancer causes hundreds of thousands deaths each year worldwide. The early stage diagnosis and treatment can significantly reduce the mortality rate. However, traditional manual needs intense workload, diagnostic errors are prone to happen with prolonged work pathologists. Automatic histopathology image recognition plays a key role in speeding up improving quality diagnosis. Methods In this work, we propose breast classification by assembling multiple compact...
The evaluation of human epidermal growth factor receptor 2 (HER2) expression is essential to formulate a precise treatment for breast cancer. routine HER2 conducted with immunohistochemical techniques (IHC), which very expensive. Therefore, the first time, we propose cancer (BCI) benchmark attempting synthesize IHC data directly paired hematoxylin and eosin (HE) stained images. dataset contains 4870 registered image pairs, covering variety levels.Based on BCI, as minor contribution, further...
In this paper, we propose a CNN based method to perform low-light image enhancement. We design special module utilize multiscale feature maps, which can avoid gradient vanishing problem as well. order preserve textures much possible, use SSIM loss train our model. The contrast of images be adaptively enhanced using method. Results demonstrate that outperforms other enhancement methods.
Deep learning-based histopathology image classification is a key technique to help physicians in improving the accuracy and promptness of cancer diagnosis. However, noisy labels are often inevitable complex manual annotation process, thus mislead training model. In this work, we introduce novel hard sample aware noise robust learning method for classification. To distinguish informative samples from harmful ones, build an easy/hard/noisy (EHN) detection model by using history. Then integrate...
Abstract Self-lubricating joint bearings play an important role in the field of aviation because they have advantageous attributes simple structures, strong load-bearing capacity and free maintenance. Fabric composite liners, as emerging frictional material for self-lubricating spherical bearings, been widely studied due to their long service life, design flexibility self-lubrication characteristics. Recently, increasing use fabric liners has promoted extensive investigation into enhancing...
In this paper, we propose a joint framework to enhance images under low-light conditions. First, convolutional neural network (CNN) based architecture is proposed denoise images. Then, on atmosphere scattering model, introduce model image contrast. our simple but effective prior, bright channel estimate the transmission parameter; besides, an filter designed adaptively environment light in different areas. Experimental results demonstrate that method achieves superior performance over other methods.
With the increasing cases of thyroid malignant tumors, diagnosis nodule has attracted more and attention. Deep learning achieved promising results in computer-aided due to advantages obtaining high-dimensional features. In this paper, we proposed a hybrid multi-branch convolutional neural network based on feature cropping method for extraction classification ultrasound images. Firstly, designed backbone extract shared maps as global branch. Next, added branch perform multi-cropping batch...
Unsupervised person re-identification (Re-ID) is a promising and very challenging research problem in computer vision. Learning robust discriminative features with unlabeled data of central importance to Re-ID. Recently, more attention has been paid unsupervised Re-ID algorithms based on clustered pseudo-label. However, the previous approaches did not fully exploit information hard samples, simply using cluster centroid or all instances for contrastive learning. In this paper, we propose...
To develop and validate a deep learning (DL)-based primary tumor biopsy signature for predicting axillary lymph node (ALN) metastasis preoperatively in early breast cancer (EBC) patients with clinically negative ALN.A total of 1,058 EBC pathologically confirmed ALN status were enrolled from May 2010 to August 2020. A DL core-needle (DL-CNB) model was built on the attention-based multiple instance-learning (AMIL) framework predict utilizing features, which extracted areas digitized...
Importance Adolescent idiopathic scoliosis (AIS) is the most common pediatric spinal disorder. Routine physical examinations by trained personnel are critical to diagnose severity and monitor curve progression in AIS. In presence of concerning malformation, radiographs necessary for diagnosis or follow-up, guiding further management, such as bracing correction moderate malformation spine surgery severe malformation. If left unattended, progressive deterioration occurs two-thirds patients,...
Abstract Inertial amplification mechanisms could be used to control the propagation of elastic waves in beams and slabs, but it was a difficult problem apply inertial seismic metamaterials design low-frequency broadband. This paper presents inertially amplified locally resonant metamaterial (IALR-SM) using coupling mechanism local resonance. In contrast (LRSM), large-mass columns as resonators IALR-SM are attached connector small-mass form inertia structures. The finite element method...
TorchAudio is an open-source audio and speech processing library built for PyTorch. It aims to accelerate the research development of technologies by providing well-designed, easy-to-use, performant PyTorch components. Its contributors routinely engage with users understand their needs fulfill them developing impactful features. Here, we survey TorchAudio's principles contents highlight key features include in its latest version (2.1): self-supervised learning pre-trained pipelines training...
Diffusion-weighted magnetic resonance imaging (DWI) is sensitive to acute ischemic stroke and a common diagnostic method for the stroke. However, result relies on visual observation of neurologists which may vary from doctor under different circumstance. And manual segmentation often time-consuming subjective process. The time onset thrombus removal has significant impact prognosis patients with shorter time, better prognosis. For this purpose we present novel framework quickly automatically...
Abstract Routine check-ups for adolescent idiopathic scoliosis are critical to monitor progression and prescribe interventions. AIS is primarily screened via physical examination. If there features of deformity, radiographs necessary diagnosis or follow-up, guiding further management, i.e., bracing moderate deformity surgery severe. However, this subjects children repetitive radiation routine practices can be disturbed. Here, we demonstrate a mobile platform powered by ScolioNet , being...
Semi-supervised domain adaptation (SSDA) has been extensively researched due to its ability improve classification performance and generalization of models by using a small amount labeled data on the target domain. However, existing methods cannot effectively adapt difficulty in fully learning rich complex semantic information relationships. In this paper, we propose novel SSDA framework called regularization (SERL), which captures from multiple perspectives achieve adaptive fine-tuning...