A5041877213- Advanced Neural Network Applications
- Video Surveillance and Tracking Methods
- Adversarial Robustness in Machine Learning
- Ethics and Social Impacts of AI
- Multimodal Machine Learning Applications
- Advanced Measurement and Detection Methods
- Privacy-Preserving Technologies in Data
- Advanced Image and Video Retrieval Techniques
- Autonomous Vehicle Technology and Safety
- Image and Object Detection Techniques
- Time Series Analysis and Forecasting
- Infrared Target Detection Methodologies
- Advanced Vision and Imaging
- Fire Detection and Safety Systems
- Infrastructure Maintenance and Monitoring
- Industrial Vision Systems and Defect Detection
- Video Analysis and Summarization
- Domain Adaptation and Few-Shot Learning
- Image Enhancement Techniques
Jingdong (China)
2020-2022
Airport Shunyi District Hospital
2021
Beijing University of Posts and Telecommunications
2015-2017
Temple University
2015-2016
Henan University of Technology
2012
Deep networks have been used for semantic segmentation tasks on scenes of outdoor environments with increasing popularity. However, the majority existing work centers daytime favorable illumination and weather conditions, relies supervision pixel-level annotations. This paper seeks to address problem rainy, night-time without using We introduce a near scene approach that uses images as bridge transferring knowledge from pre-trained models rainy night images. Specifically, we first present...
Understanding traffic scene images taken from vehicle mounted cameras is important for high-level tasks, such as advanced driver assistance systems and autonomous driving. It a challenging problem due to large variations under different weather or illumination conditions. In this paper, we tackle the of understanding cross-domain perspective. We attempt understand same location but conditions (e.g., on rainy night with help sunny day). To end, propose dense correspondence-based transfer...
The rapid development of Artificial Intelligence (AI) technology has enabled the deployment various systems based on it. However, many current AI are found vulnerable to imperceptible attacks, biased against underrepresented groups, lacking in user privacy protection. These shortcomings degrade experience and erode people's trust all systems. In this review, we provide practitioners with a comprehensive guide for building trustworthy We first introduce theoretical framework important aspects...
In order to improve accuracy and robustness of the lane detection in complex conditions, such as shadows illumination changing, a novel algorithm was proposed based on machine learning. After pretreatment, set haar-like filters were used calculate eigenvalue gray image f(x,y) edge e(x,y). Then these features trained by using improved boosting final class function g(x) obtained, which judge whether point x belonging or not. To avoid over fitting traditional boosting, Fisher discriminant...
Understanding traffic scene images taken from vehicle mounted cameras provides important information for high level tasks such as autonomous driving and advanced driver assistance. The problem is hard due to challenges weather illumination variation. To facilitate the research against challenges, in this paper we present a new benchmark cross-weather understanding <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> . dataset consists of...
Parsing road scene images taken from vehicle mounted cameras provides important information for high level tasks in automated on-road vehicles. In this paper we adopt the nonparametric framework problem and present a simple yet effective strategy to integrate spatial prior into framework. Unlike natural images, our typically have very stable layout, which motivates us explore such layout improving labeling. particular, distribution of each semantic category is obtained set previously...
In order to improve the adaptability of lane detection algorithm under complex conditions such as damaged lines, covered shadow, insufficient light, rainy day etc. Lane based on Zoning Hough Transform is proposed in this paper. The road images are processed by improved ±45° Sobel operators and two-dimension Otsu algorithm. To eliminate interference ambient noise, highlight dominant position lane, used, which can obtain parameters identify accurately. experiment results show method extract...
This paper studies and designs digital image recognition algorithms based on pattern recognition. The feature vector includes four-dimensional vector, eight-dimensional two-dimensional principal component analysis. classification methods include K-nearest neighbor method, minimum distance method fixed increment method. Through the combination of different vectors methods, results are obtained. At same time, advantages, disadvantages accuracy three compared. show that has high accuracy, it is...
Intelligent evaluation of carbon emission reduction level in thermal
In the apron target detection, size is very small, and a large number of fine particle characteristics are lost in compression process, resulting recognition errors. Fine granularity directly affects accuracy recognition. However, due to limited energy consumption, computational power equipment, insufficient feature extraction small targets by universal detection algorithms, improvement speed such algorithms restricted. This paper presents real-time algorithm which named DT-YOLOV4 (difficult...