A5009482239- Autonomous Vehicle Technology and Safety
- Video Surveillance and Tracking Methods
- Cutaneous Melanoma Detection and Management
- Traffic Prediction and Management Techniques
- 3D Surveying and Cultural Heritage
- Robotic Path Planning Algorithms
- 3D Shape Modeling and Analysis
- AI in cancer detection
- Simulation Techniques and Applications
- Robotics and Sensor-Based Localization
- Advanced Measurement and Detection Methods
- Time Series Analysis and Forecasting
- Computer Graphics and Visualization Techniques
- Human Pose and Action Recognition
- Systemic Sclerosis and Related Diseases
- Genetic and rare skin diseases.
- melanin and skin pigmentation
- Infrared Target Detection Methodologies
- Advanced Neural Network Applications
- Vehicular Ad Hoc Networks (VANETs)
- Nonmelanoma Skin Cancer Studies
Beihang University
2017-2022
Megvii (China)
2019-2020
Predicting the future behavior of agents is a fundamental task in autonomous vehicle domains. Accurate prediction relies on comprehending surrounding map, which significantly regularizes agent behaviors. However, existing methods have limitations exploiting map and exhibit strong dependence historical trajectories, yield unsatisfactory performance robustness. Additionally, their heavy network architectures impede real-time applications. To tackle these problems, we propose Map-Agent Coupled...
Skin diseases are very common in our daily life. Due to the similar appearance of skin diseases, automatic classification through lesion images is quite a challenging task. In this paper, novel multi-classification method based on convolutional neural network (CNN) proposed for dermoscopy images. A CNN with nested residual structure designed first, which can learn more information than original structure. Then, trained transfer learning. With network, 6 kinds classified, including nevus,...
This paper introduces MCTrack, a new 3D multi-object tracking method that achieves state-of-the-art (SOTA) performance across KITTI, nuScenes, and Waymo datasets. Addressing the gap in existing paradigms, which often perform well on specific datasets but lack generalizability, MCTrack offers unified solution. Additionally, we have standardized format of perceptual results various datasets, termed BaseVersion, facilitating researchers field (MOT) to concentrate core algorithmic development...
This technical report presents an effective method for motion prediction in autonomous driving. We develop a Transformer-based input encoding and trajectory prediction. Besides, we propose the Temporal Flow Header to enhance encoding. In end, efficient K-means ensemble is used. Using our Transformer network method, win first place of Argoverse 2 Motion Forecasting Challenge with state-of-the-art brier-minFDE score 1.90.
End-to-End paradigms use a unified framework to implement multi-tasks in an autonomous driving system. Despite simplicity and clarity, the performance of end-to-end methods on sub-tasks is still far behind single-task methods. Meanwhile, widely used dense BEV features previous make it costly extend more modalities or tasks. In this paper, we propose Sparse query-centric paradigm for Autonomous Driving (SparseAD), where sparse queries completely represent whole scenario across space, time...
In this paper, we propose a novel learning-based pipeline for partially overlapping 3D point cloud registration. The proposed model includes an iterative distance-aware similarity matrix convolution module to incorporate information from both the feature and Euclidean space into pairwise matching process. These layers learn match points based on joint of entire geometric features offset each pair, overcoming disadvantage by simply taking inner product vectors. Furthermore, two-stage...
Objective To compare the diagnostic accuracies of deep convolutional neural network (CNN) and dermatologists for pigmented nevus seborrheic keratosis. Methods CNN ResNet-50 was trained with 5 094 dermoscopic images keratosis using transfer learning, so as to establish a CNN two-classification model. Then, this model applied automatic classification 30 keratosis. Meanwhile, in combination clinical photos skin lesions, 95 experienced who had received dermoscopy training gave their...
3D multi-object tracking and trajectory prediction are two crucial modules in autonomous driving systems. Generally, the tasks handled separately traditional paradigms a few methods have started to explore modeling these joint manner recently. However, approaches suffer from limitations of single-frame training inconsistent coordinate representations between tasks. In this paper, we propose streaming unified framework for Multi-Object Tracking Prediction (StreamMOTP) address above...
Trajectory generation is a pivotal task in autonomous driving. Recent studies have introduced the autoregressive paradigm, leveraging state transition model to approximate future trajectory distributions. This paradigm closely mirrors real-world process and has achieved notable success. However, its potential limited by ineffective representation of realistic trajectories within redundant space. To address this limitation, we propose Kinematic-Driven Generative Model for Realistic Agent...
Predicting the future behavior of agents is a fundamental task in autonomous vehicle domains. Accurate prediction relies on comprehending surrounding map, which significantly regularizes agent behaviors. However, existing methods have limitations exploiting map and exhibit strong dependence historical trajectories, yield unsatisfactory performance robustness. Additionally, their heavy network architectures impede real-time applications. To tackle these problems, we propose Map-Agent Coupled...
The computer-aided diagnosis (CAD) system can provide a reference basis for the clinical of skin diseases. Convolutional neural networks (CNNs) not only extract visual elements such as colors and shapes but also semantic features. As they have made great improvements in many tasks dermoscopy images. imaging has no principal orientation, indicating that there are large number lesion rotations datasets. However, CNNs lack rotation invariance, which is bound to affect robustness against...