A5041316314- Cellular Mechanics and Interactions
- Advanced Vision and Imaging
- Cancer Cells and Metastasis
- 3D Printing in Biomedical Research
- Gaze Tracking and Assistive Technology
- Human Pose and Action Recognition
- Anomaly Detection Techniques and Applications
- COVID-19 diagnosis using AI
- Image Processing Techniques and Applications
- Soft Robotics and Applications
- Video Surveillance and Tracking Methods
- COVID-19 Clinical Research Studies
- Microfluidic and Bio-sensing Technologies
- Drilling and Well Engineering
- COVID-19 epidemiological studies
- Force Microscopy Techniques and Applications
- SARS-CoV-2 and COVID-19 Research
- Cardiac Valve Diseases and Treatments
- CCD and CMOS Imaging Sensors
- Ocular Disorders and Treatments
- Chemical Reactions and Isotopes
- Infective Endocarditis Diagnosis and Management
- Robot Manipulation and Learning
- thermodynamics and calorimetric analyses
- Advanced Image Processing Techniques
Tsinghua–Berkeley Shenzhen Institute
2020-2024
Tsinghua University
2020-2024
We present PANDA, the first gigaPixel-level humAN-centric viDeo dAtaset, for large-scale, long-term, and multi-object visual analysis. The videos in PANDA were captured by a gigapixel camera cover real-world scenes with both wide field-of-view (~1 square kilometer area) high-resolution details (~gigapixel-level/frame). may contain 4k head counts over 100× scale variation. provides enriched hierarchical ground-truth annotations, including 15,974.6k bounding boxes, 111.8k fine-grained...
Semantic understanding of 3D environments is critical for both the unmanned system and human involved virtual/augmented reality (VR/AR) immersive experience. Spatially-sparse convolution, taking advantage intrinsic sparsity point cloud data, makes high resolution convolutional neural networks tractable with state-of-the-art results on semantic segmentation problems. However, exhaustive computations limits practical usage perception VR/AR applications in portable devices. In this paper, we...
The ability of camera arrays to efficiently capture higher space-bandwidth product than single cameras has led various multiscale and hybrid systems. These systems play vital roles in computational photography, including light field imaging, 360 VR camera, gigapixel videography, etc. One the critical tasks imaging is matching fusing cross-resolution images from different under perspective parallax. In this paper, we investigate reference-based super-resolution (RefSR) problem associated with...
The emergence of the novel coronavirus disease 2019 (COVID-19) is placing an increasing burden on healthcare systems. Although majority infected patients experience non-severe symptoms and can be managed at home, some individuals develop severe require hospital admission. Therefore, it critical to efficiently assess severity COVID-19 identify hospitalization priority with precision. In this respect, a four-variable assessment model, including lymphocyte, lactate dehydrogenase, C-reactive...
Abstract Cable-driven soft continuum robots are important tools in minimally invasive surgery (MIS) to reduce the lesions, pain and risk of infection. The feasibility employing deep reinforcement learning (DRL) for controlling cable-driven has been investigated; however, a considerable gap between simulations real world exists. Here we introduce learning-based method, Axis-Space (AS) framework, which accelerates computational speed improves accuracy robotic control by reducing sample...
Abstract Tumor spread is responsible for most deaths related to cancer. Increasing the accuracy of cancer prognosis critical reducing high mortality rates in patients. Here, we report that electrostatic potential difference (EPD) between tumor and its paratumor tissue a prognostic marker spread. This finding concluded from patient‐specific EPD values clinical observation. The were measured on cryosections 51 patients using Kelvin probe force microscopy (KPFM). A total ~44% (15/34) V...
(Patterns 1, 100092-1--100092-9; October 9, 2020) In the originally published version of this article, formulas for Sensitivity, Specificity, and F1 weighted in Table S3, shown below, were incorrect.Tabled 1MetricsDefinitionSensitivityTPTP+FPSpecificityTPTN+FPF1 weighted1TP+FP∗2TPTP+FN+1+1TN+FN∗2TNTN+FP+1 Open table a new tab The correct are:Tabled 1MetricsDefinitionSensitivityTPTP+FNSpecificityTNTN+FPF1 weighted2TP2TP+FN+FP∗TP+FNTP+FN+FP+TN+2TN2TN+FP+FN∗TN+FPTP+FN+FP+TN article has been...
Automated segmentation and evaluation algorithms have been demonstrated to enhance the simplicity translational utility of organoid technology. However, there is a pressing need for development complex organoids that possess epithelium environmental elements, dense regional cell aggregation, intraorganoid morphologies. Nevertheless, has limited progress, including both construction data sets algorithms, in use user-friendly microscopy address such organoids. In this study, set bright-field...
Pupil tracking is an important branch of object which require high precision. We investigate head mounted pupil often more convenient and precise than remote tracking, but also challenging. When suffers from noise like bad illumination, detection precision dramatically decreases. Due to the appearance recording device public benchmark image datasets, algorithms have become easier design evaluate. In this paper, we propose a robust algorithm uses Convolutional Neural Network (CNN) combine...
We present PANDA, the first gigaPixel-level humAN-centric viDeo dAtaset, for large-scale, long-term, and multi-object visual analysis. The videos in PANDA were captured by a gigapixel camera cover real-world scenes with both wide field-of-view (~1 square kilometer area) high-resolution details (~gigapixel-level/frame). may contain 4k head counts over 100x scale variation. provides enriched hierarchical ground-truth annotations, including 15,974.6k bounding boxes, 111.8k fine-grained...
Sensing and understanding large-scale dynamic scenes require a high-performance imaging system. Conventional systems pursue higher capability by simply increasing the pixel resolution via stitching cameras at expense of bulky Moreover, they strictly follow feedforward pathway: that is, their pixel-level sensing is independent semantic understanding. Differently, human visual system owns superiority with both feedback pathways: The pathway extracts object representation (referred to as memory...
Tumor spread is responsible for most deaths related to cancer. Increasing the accuracy of cancer prognosis could be key reduce high mortality rates in patients. Here, we report that electrostatic potential difference between tumor and paratumor tissue cryosections serves as a prognostic marker spread. We demonstrated polarity this correlated with patients abundance stem cells tissues. Then, using spheroids mimic microenvironment, showed selectively migrate toward negatively charged...
Abstract Soft continuum robots undergoes nonlinear spatial deformation. Deep reinforcement learning (DRL) suffers from large training dataset and high time consumption. This paper reports a generalized principle, N-space (NS) framework, that employs on value-based algorithms (e.g. DQN DDQN), to overcome the challenge of local minima associated with sample efficiency or limited consumption in online training. The performance NS-augmented DRL (NS-DRL) was examined controlling self-design...