Taipei City, Taiwan
As an AI engineer with a strong background in machine learning and computer science, I am excited to bring my skills and experience to bear on challenging and impactful projects. With a proven track record of developing and implementing innovative AI solutions, I am confident in my ability to drive results and make a meaningful contribution to any team. Whether developing new algorithms, optimizing existing models, or working with large and complex data sets, I thrive on the opportunity to apply my skills and expertise to real-world problems. I am passionate about staying up-to-date with the latest developments in the field and am always looking for ways to expand my knowledge and improve my abilities.
- Lead discussion sections and homework tutorials.
- Evaluate and grade assignments, papers and record grades.
- Follow established guidelines and gather relevant data for projects and reports.
April 2018 - July 2018
- Lead discussion sections and homework tutorials.
- Evaluate and grade assignments, papers and record grades.
- Follow established guidelines and gather relevant data for projects and reports.
May 2019 - Jun 2019
September 2020 - Jan 2021
2015 - 2020
2021 - 2023
TY6B4QCK387W
Issued December 2020 · No Expiration Date
19VN000307VOH276A
Issued May 2019 · Expired May 2021
2016 - 2020
May 2016
April 2018
May 2018
May 2019
2021 - 2023
This work presents a new approach to 360° video saliency prediction using a spherical convolutional network. The proposed method focuses on defining the convolutional kernel as a spherical cap. During the convolution process, instead of using neighboring pixels with regular relationships in the equirectangular projection coordinate, the convolutional patches are transformed to maintain the spherical perspective of the spherical signal.
This work introduces an enhanced version of RandLA-Net, named RandLASAMP-Net, a neural network designed for efficient per-point semantics inference on large-scale point clouds. Despite its computational and memory efficiency, the local feature aggregation modules in RandLA-Net still rely on a basic attention mechanism. To enhance the performance of the local feature aggregation module in point cloud segmentation tasks, the authors propose integrating a self-attention mechanism. Moreover, a multi-pyramid module is utilized to increase the ability to learn global and local context feature.
Taipei City, Taiwan
As an AI engineer with a strong background in machine learning and computer science, I am excited to bring my skills and experience to bear on challenging and impactful projects. With a proven track record of developing and implementing innovative AI solutions, I am confident in my ability to drive results and make a meaningful contribution to any team. Whether developing new algorithms, optimizing existing models, or working with large and complex data sets, I thrive on the opportunity to apply my skills and expertise to real-world problems. I am passionate about staying up-to-date with the latest developments in the field and am always looking for ways to expand my knowledge and improve my abilities.
- Lead discussion sections and homework tutorials.
- Evaluate and grade assignments, papers and record grades.
- Follow established guidelines and gather relevant data for projects and reports.
April 2018 - July 2018
- Lead discussion sections and homework tutorials.
- Evaluate and grade assignments, papers and record grades.
- Follow established guidelines and gather relevant data for projects and reports.
May 2019 - Jun 2019
September 2020 - Jan 2021
2015 - 2020
2021 - 2023
TY6B4QCK387W
Issued December 2020 · No Expiration Date
19VN000307VOH276A
Issued May 2019 · Expired May 2021
2016 - 2020
May 2016
April 2018
May 2018
May 2019
2021 - 2023
This work presents a new approach to 360° video saliency prediction using a spherical convolutional network. The proposed method focuses on defining the convolutional kernel as a spherical cap. During the convolution process, instead of using neighboring pixels with regular relationships in the equirectangular projection coordinate, the convolutional patches are transformed to maintain the spherical perspective of the spherical signal.
This work introduces an enhanced version of RandLA-Net, named RandLASAMP-Net, a neural network designed for efficient per-point semantics inference on large-scale point clouds. Despite its computational and memory efficiency, the local feature aggregation modules in RandLA-Net still rely on a basic attention mechanism. To enhance the performance of the local feature aggregation module in point cloud segmentation tasks, the authors propose integrating a self-attention mechanism. Moreover, a multi-pyramid module is utilized to increase the ability to learn global and local context feature.