Pattaramanee Arsomngern (Ploy)

ภัทรมณี อาศรมเงิน (พลอย)

Ph.D. student, School of Information Science and Technology (VISTEC), Thailand

E-mail: pattaramanee.a_s19 at vistec dot ac dot th

GitHub | Google Scholar | DBLP | Linkedin

CV

About

I am currently a 4th year CS Ph.D. candidate from the School of Information Science and Technology (VISTEC), Thailand. I am under the supervision of Prof. Supasorn Suwajanakorn as a part of members in Vision & Learning Lab (VLL) and Prof. Sarana Nutanong as a part of members in SCAlable Data System lab (SCADS) and a team leader of Retrieval & Representation Learning research group. I also collaborate with Asst. Prof. Dr. Cheng Long from Nanyang Technological University (NTU), Singapore. I was granted 27,000 USD Thai royal scholarship to conduct research overseas, thus I am seeking for a research team to join as an intern or a visiting Ph.D. student in 2023-2024.

Bio

I received the B.Eng. degree in Computer engineering (Top 5% of a class, first-class honors) from KMUTT, Thailand in 2019. My undergrad’s final project is about anomaly detection on animal medical images, under the supervision of Prof. Phond Phunchongharn. I did an internship during undergrad as a research assistant at Computer Networks and Communications Lab (CNCL)@NTU, and Chulabhorn Graduate Institute, Thailand.

Research

My research interests are about representation learning in computer vision (i.e., images and point clouds), graphs, and sets. Specifically, I have extensive experience in designing pre-training objectives for given specific downstream tasks (e.g., semantic segmentation, object detection, classification, or retrieval).

Peer-reviewed articles

• Learning Geometric-aware Properties in 2D Representation Using Lightweight CAD Models, or Zero Real 3D Pairs

  P Arsomngern, S Nutanong, S Suwajanakorn

  In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023 [paper] [project page] (CORE A*)

  • Introducing a new way to learn a visual representation for better solving 2D Indoor Scene understanding tasks (i.e., semantic segmentation, object detection, and instance segmentation). This project introduces a novel technique to pre-train such a model that could given discriminative and useful visual and geometric prior for our desired downstream task without relying on pre-training on limited 3D indoor scenes as in prior works.

• Towards Pointsets Representation Learning via Self-Supervised Learning and Set Augmentation

  P Arsomngern, C Long, S Suwajanakorn, S Nutanong

  In IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2022 [paper] [code] (IF: 24.314)

• Self-Supervised Deep Metric Learning for Pointsets

  P Arsomngern, C Long, S Suwajanakorn, S Nutanong

  In IEEE International Conference on Data Engineering (ICDE), 2021 [paper] [code] (CORE A*)

• Computer-aided diagnosis for lung lesion in companion animals from x-ray images using deep learning techniques

  P Arsomngern, N Numcharoenpinij, J Piriyataravet, W Teerapan, W Hinthong, P Phunchongharn

  In IEEE International Conference on Awareness Science and Technology (iCAST), 2019 [paper]

  • Pet-X got 2nd prize (without first prize) at Thailand National Software Contest (NSC), 2019, under the “Applications for science and technologies” category.

Ongoing projects

• Heterogeneous Graph Representation Learning

  (P Arsomngern as a co-author)

  • Exploring an alternative method to self-discrimination learning the representation based on augmented heterogeneous graph structures.

• Deep Fake Video Retrieval

  (P Arsomngern as a co-author)

Academic Services

• Journal invited reviewer: IEEE TPAMI

Technical Skills

Programming languages

• Python (professional)
• SQL, NoSQL, JavaScript, CSS, PHP, Java, C#, R (limited)

Frameworks, tools and libraries

• PyTorch, Tensorflow2.X, Keras, PyTorch geometric, Tensorboard, matplotlib, numpy, scikit-learn, git, Linux (professional)
• Node.js, React.js, Unity, Google Cloud Platform (limited)

Languages

• Thai (Native)
• English (CU-TEP 83/120 (TOEFL iBT equivalent, CEFR B2))
• Japanese (JLPT N2)