Sherzod

Online Resume

Download PDF Version

Sherzod Salokhiddinov

AI & Computer Vision Engineer
image

Experienced AI specialist with a strong focus on 3D technologies and computer vision, driving innovation across various industries. Currently leading projects at Riso Convergence, where I manage the development of AI-driven solutions, including 3D reconstruction, generative 3D, and image processing for devices like scanners and fax machines. My expertise spans multiple domains, from creating web-based 3D viewers to optimizing AI models for performance and accuracy.

AI • Computer Vision • 3D Reconstruction • Neural Radiance Fields (NeRF) • Generative-3D • Deep Learning

Work Experiences

Senior AI Research Engineer

Riso Convergence | 2023.11 - Present

Driving innovation through AI and 3D technologies across multiple projects.

  • Scanner/Fax Device Control API: Developed backend components for a DataMatrix decoding project, including controlling scanner/fax devices, receiving documents, and applying image processing techniques to accurately read DataMatrix codes.
  • SupperCaddy App: Created dynamic scorecard content by leveraging AI-based depth effect animations generated from a single image.
  • 3D-Viewer & Converter: Created a web-based 3D viewer using Three.js, incorporating advanced 3D controls such as tree-based layer views, clipping planes, and various camera types. Additionally, managed the 3D model conversion process for multiple formats, including JT, FBX, GLB/GLTF, and OBJ.

Senior AI Research Engineer

FOURLAB (포랩) | 2022.09 - 2023.10

Leading AI-driven projects with a focus on 3D technologies and innovation.

  • Pix2Poly App: Managed the Pix2Poly project, overseeing the development of 3D processing components, including 3D reconstruction, Generative 3D (from single image and text), 3D file format conversion, Blender plugin development, and text-to-3D effects.
  • Led NeRF-based 3D reconstruction projects aimed at developing Metaverse and Digital Twin applications, utilizing tools such as NeRF, Three.js, Flask, and PyTorch.
  • Demo Videos:
  • Technologies: Photogrammetry, NeRF (Neural Radiance Fields), 3D Gaussian Splatting, etc.

AI & Computer Vision Engineer

Far Island Corp. | 2021.03 - 2022.08

Developed advanced AI algorithms for Machine Vision applications.

  • Deep Learning Framework: Developed AI components for a deep learning framework with a GUI tailored for inspection systems, optimizing machine vision tasks.
  • Bolt Inspection System: Designed and deployed an image classification algorithm for bolt inspection, achieving 94% accuracy, significantly improving the quality control process.
  • Implemented object detection and image classification models in PyTorch, and optimized inference speed by converting to TensorRT, resulting in a sixfold increase in performance
  • Synthetic Dataset Creation: Created synthetic datasets using Blender and Python scripts to improve the diversity and robustness of training data, leading to more accurate and reliable model outputs.
  • 3D Reconstruction and Point Cloud Registration: Developed advanced techniques for 3D reconstruction and point cloud registration using multi-view images, enhancing the company's capabilities in 3D modeling and analysis.
  • Collaborated with Java experts to integrate AI-driven controls into a project managing the xArm-6 robot arm via TCP/IP, utilizing the OAK-D camera and DepthAI library for precise automation.

Research Assistant, Perception & Computer Vision Lab.

Kyung Hee University | 2014 - 2021
  • Conducted advanced research in 3D Reconstruction, Depth from Focus, and Face Detection utilizing both AI-based and conventional methods, contributing to the academic field of knowledge with several peer-reviewed publications.
  • Led the development of a 3D model reconstruction process for dental structures from medical imaging devices, employing parallel programming with OpenMP in C/C++ to improve processing efficiency.
  • Developed and implemented depth estimation methods using 3D Convolutional Neural Networks (3D-CNN) and hybrid CNN + LSTM architectures, advancing depth perception technologies.

Publications

  • Salokhiddinov Sherzod and Seungkyu Lee. “Deep Spatio-focal Network for Depth from Focus”. Journal of Imaging Science and Technology (2021).
  • Salokhiddinov Sherzod and Seungkyu Lee. "Iterative Refinement of Uniformly Focused Image Set for Accurate Depth from Focus." Applied Sciences 10.23 (2020): 8522.
  • Salokhiddinov Sherzod and Seungkyu Lee. "Depth from focus for 3D reconstruction by iteratively building uniformly focused image set." ACM SIGGRAPH 2018 Posters. 2018. 1-2.
  • Salokhiddinov Sherzod and Seungkyu Lee. "3D Surface Reconstruction Using Sequence of Images Based on Depth from Defocus Technique." Journal of Korean Information Science Society (2016): 1114-1116.
  • Salokhiddinov Sherzod and Seungkyu Lee. "Small Noisy and Perspective Face Detection using Deformable Symmetric Gabor Wavelet Network" arXiv (2020).