Yi Lu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Yi Lu has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
- Jiangtian Pan
Megvii-inc, China
1 shared publication - Jing Cao
Northeastern University, USA
1 shared publication - Muyun Li
Independent Researcher, China
1 shared publication - Qiyuan Tian
George Washington University, USA
1 shared publication - Xianting Wu
Columbia University, USA
1 shared publication - You Yao
USC Viterbi School of Engineering, University of Southern California, Los Angeles, CA, USA
1 shared publication
Latest Publications
- Open Access
- 27 November 2024
- Published by MDPI AG in Preprints.org
Advanced Deep Learning for Precise Multi-Organ Functional Tissue Segmentation
This study aims to advance medical research by identifying and segmenting functional tissue units (FTUs) within five human organs using deep learning techniques. The dataset comprises tissue slice images from the Human Protein Atlas (HPA) and the Human BioMolecular Atlas Program (HuBMAP). We assess segmentation accuracy using the mean Dice coefficient. Analysis indicates significant distributional differences across gender and age groups, prompting the design of varied sample weighting coefficients and sampling and embedding strategies. The proposed model, SegF x 2 + EffB7-B8 + FPN + ASPP, demonstrated superior performance compared to other models. This paper discusses data preprocessing, model definition, and integration, showcasing how our approach surpasses existing methods.
- 27 November 2024
- Published by MDPI AG in Preprints.org
Open AccessAdvanced Deep Learning for Precise Multi-Organ Functional Tissue Segmentation
This study aims to advance medical research by identifying and segmenting functional tissue units (FTUs) within five human organs using deep learning techniques. The dataset comprises tissue slice images from the Human Protein Atlas (HPA) and the Human BioMolecular Atlas Program (HuBMAP). We assess segmentation accuracy using the mean Dice coefficient. Analysis indicates significant distributional differences across gender and age groups, prompting the design of varied sample weighting coefficients and sampling and embedding strategies. The proposed model, SegF x 2 + EffB7-B8 + FPN + ASPP, demonstrated superior performance compared to other models. This paper discusses data preprocessing, model definition, and integration, showcasing how our approach surpasses existing methods.
- Open Access
- 15 November 2024
- Published by MDPI AG in Preprints.org
Enhancing Autonomous Vehicle Perception: A 6-DOF Pose Estimation Framework Using Deep Learning
This research addresses the critical challenge of precise perception of surrounding vehicles' positions and orientations in autonomous driving environments. We aim to develop an algorithm to estimate the absolute 6-degree-of-freedom (6-DOF) pose of vehicles from single images using a dataset of over 60,000 labeled 3D vehicle instances. Our approach integrates Faster R-CNN for 2D bounding box detection, Q-Net for 6D-pose regression, and DenseNet for enhanced feature extraction. Data analysis revealed attribution errors in pitch and yaw and significant variance in spatial accuracy across different vehicle models. By fine-tuning the Faster R-CNN on BDD100k and our dataset, and employing weighted bounding box and point fusion techniques, our model outperforms existing methods in accuracy and robustness. This work provides a reliable method for autonomous vehicles to perceive their surroundings accurately, enhancing navigation safety and efficiency.
- 15 November 2024
- Published by MDPI AG in Preprints.org
Open AccessEnhancing Autonomous Vehicle Perception: A 6-DOF Pose Estimation Framework Using Deep Learning
This research addresses the critical challenge of precise perception of surrounding vehicles' positions and orientations in autonomous driving environments. We aim to develop an algorithm to estimate the absolute 6-degree-of-freedom (6-DOF) pose of vehicles from single images using a dataset of over 60,000 labeled 3D vehicle instances. Our approach integrates Faster R-CNN for 2D bounding box detection, Q-Net for 6D-pose regression, and DenseNet for enhanced feature extraction. Data analysis revealed attribution errors in pitch and yaw and significant variance in spatial accuracy across different vehicle models. By fine-tuning the Faster R-CNN on BDD100k and our dataset, and employing weighted bounding box and point fusion techniques, our model outperforms existing methods in accuracy and robustness. This work provides a reliable method for autonomous vehicles to perceive their surroundings accurately, enhancing navigation safety and efficiency.