A5115603705- Advanced Image Processing Techniques
- Image Enhancement Techniques
- Advanced Vision and Imaging
- Image Retrieval and Classification Techniques
- Generative Adversarial Networks and Image Synthesis
- Visual Attention and Saliency Detection
- Vehicle License Plate Recognition
- Image and Signal Denoising Methods
- Face recognition and analysis
- Advanced Image and Video Retrieval Techniques
- Advanced Neural Network Applications
- Chaos-based Image/Signal Encryption
- Genetics and Neurodevelopmental Disorders
- Multimodal Machine Learning Applications
- Air Traffic Management and Optimization
- Human Pose and Action Recognition
- Parallel Computing and Optimization Techniques
- Smart Agriculture and AI
- Video Surveillance and Tracking Methods
- Video Analysis and Summarization
- Biometric Identification and Security
- Computer Science and Engineering
- Recommender Systems and Techniques
- Handwritten Text Recognition Techniques
- Hydrology and Sediment Transport Processes
Southeast University
2025
South China Normal University
2024
Civil Aviation University of China
2024
China Southern Power Grid (China)
2024
Harbin Institute of Technology
2020-2023
Institute of Soil and Water Conservation
2022
Northwest A&F University
2022
Minzu University of China
2021
Xidian University
2021
Harbin Engineering University
2018
Person re-identification (Re-ID) is a fundamental subject in the field of computer vision technologies. The traditional methods person Re-ID have difficulty solving problems illumination, occlusion and attitude change under complex background. Meanwhile, introduction deep learning opens new way research becomes hot spot this field. This study reviews Re-ID, then authors focus on related papers about different frameworks basis learning, discusses their advantages disadvantages. Finally, they...
Learning RAW-to-sRGB mapping has drawn increasing attention in recent years, wherein an input raw image is trained to imitate the target sRGB captured by another camera. However, severe color inconsistency makes it very challenging generate well-aligned training pairs of and images. While learning with inaccurately aligned supervision prone causing pixel shift producing blurry results. In this paper, we circumvent such issue presenting a joint model for alignment mapping. To diminish effect...
Usher syndrome (USH) is a rare genetic disease characterized by sensorineural deafness and blindness called retinitis pigmentosa, it inherited in an autosomal recessive pattern with prevalence of four to 17 per 100,000 people worldwide. In this study, consanguineous Chinese family USH, including two affected individuals five unaffected individuals, was recruited. All subjects received ophthalmic examination auditory examination. The USH patients exhibited severe early‐onset hearing vision...
This paper reviews the second AIM learned ISP challenge and provides description of proposed solutions results. The participating teams were solving a real-world RAW-to-RGB mapping problem, where to goal was map original low-quality RAW images captured by Huawei P20 device same photos obtained with Canon 5D DSLR camera. considered task embraced number complex computer vision subtasks, such as image demosaicing, denoising, white balancing, color contrast correction, demoireing, etc. target...
The rapid advancement of unmanned aerial vehicle (UAV) technologies has led to an increasing demand for UAV operations in low-altitude, high-density, and complex airspace such as mountains or urban areas. In order handle scenarios ensure flight safety UAVs with different missions beyond visual line sight environments, a fusion framework onboard autonomous trajectory management decision-making system using global strategical path planning local tactical optimization combination is proposed...
Autonomous underwater vehicle (AUV) has many intelligent optical system, which can collect signal information to make the system decision. One of them is vision and it capture images analyze. The performance particle image segmentation plays an important role in monitoring mineral resources. In order improve performance, some novel algorithm architectures are proposed. this paper, improved proposed based on modified U-Net. pyramid upsampling module residual bring into U-Net model, called...
Image retouching, aiming to regenerate the visually pleasing renditions of given images, is a subjective task where users are with different aesthetic sensations. Most existing methods adopt deterministic model learn retouching style from specific expert, making it less flexible meet diverse preferences. Besides, intrinsic diversity an expert due targeted processing images also deficiently described. To circumvent such issues, we propose image normalizing flow-based architectures. Unlike...
Despite recent advances in deep learning-based face frontalization methods, photo-realistic and illumination preserving frontal synthesis is still challenging due to large pose discrepancy during training. We propose a novel Flow-based Feature Warping Model (FFWM) which can learn synthesize images with inconsistent supervision. Specifically, an Illumination Preserving Module (IPM) proposed image from pairs. IPM includes two pathways collaborate ensure the synthesized are fine details....
Incident scene text detection, as the most crucial step of an incident recognition system, has received increasing research attention. In this paper, a two-layer mobile federated learning model (TMFL) is proposed to protect data privacy and improve training efficiency. Particularly, fast detector detect multi-directional multi-scale by using asymmetric convolution based feature pyramid network (AC-FPN). Compared with traditional pyramid, convolutions can effectively extract...
Existing unpaired low-light image enhancement approaches prefer to employ the two-way GAN framework, in which two CNN generators are deployed for and degradation separately. However, such data-driven models ignore inherent characteristics of transformation between low normal light images, leading unstable training artifacts. Here, we propose leverage invertible network enhance forward process degrade normal-light one inversely with learning. The generated real images then fed into...
Image retouching, aiming to regenerate the visually pleasing renditions of given images, is a subjective task where users are with different aesthetic sensations. Most existing methods deploy deterministic model learn retouching style from specific expert, making it less flexible meet diverse preferences. Besides, intrinsic diversity an expert due targeted processing on images also deficiently described. To circumvent such issues, we propose image normalizing flow-based architectures. Unlike...
Learning RAW-to-sRGB mapping has drawn increasing attention in recent years, wherein an input raw image is trained to imitate the target sRGB captured by another camera. However, severe color inconsistency makes it very challenging generate well-aligned training pairs of and images. While learning with inaccurately aligned supervision prone causing pixel shift producing blurry results. In this paper, we circumvent such issue presenting a joint model for alignment mapping. To diminish effect...