Hello, I'm Jialin Yu

I have 5 years working experience in industry including 3 years as a machine vision software engineer at TAILIN Bioengineering. My position has engaged in the domain of computer vision, image processing and biomedical imaging. I received my MSc in Computer Vison from the [University College London] in 2019 where I was supervised by Lewis Griffin.


News

  • Apr 2023: I collaborated with Sixth Tone to produce the article ‘Looking for Closure, a Grandson Built a Ghost in the Machine’.
  • Jun 2022: I was interviewed by Southern Weekly (南方周末), where my story and research were further highlighted.
  • May 2022: My story led to an interviewed and shared with Sanlian Lifeweek (三联生活周刊).
  • Apr 2022: I developed an AI-driven system integrating language models, text-to-speech (TTS), and facial reenactment to create a virtual avatar of my late grandfather, offering a heartfelt farewell that was missed on the day of his passing. This work was featured on Guokr News.

Manuscripts

Colonoscopy polyp detection with massive endoscopic images

Colonoscopy polyp detection with massive endoscopic images

Jialin Yu, Huogen Wang, Ming Chen
arXiv, 2022

Using an optimized RetinaNet structure for polyp detection.

Projects

Hover Image ChromaPH

ChromaPH

ChromaPH is an image-processing algorithm that determines pH values from a single photo. By automatically identifying and comparing the color regions on a pH test strip and its reference chart, it provides fast and accurate measurements.

ColonyTrack

ColonyTrack

ColonyTrack is an image-processing algorithm for automated colony counting and growth analysis. By aligning images of the same petri dish taken at different time points, it tracks colony development and provides accurate growth-based quantification.

ColonyTrack-Contact plate

ColonyTrack-Contact plate

An algorithm for automated colony counting on contact plates. It enhances noise resistance by handling grid lines and logos while aligning multi-timepoint images to track colony growth with high accuracy.

Cardiomyocytes Contraction Analysis

Cardiomyocytes Contraction Analysis

By processing in vitro heart cell videos, the algorithm accurately detects and quantifies rhythmic contractions, providing crucial insights for cardiac research and therapeutic development.

Hover Image Ultrafine Standardized Card Object Detection

Ultrafine Standardized Card Object Detection

An algorithm for detecting targets on standardized test cards. Capable of identifying features as small as 0.13mm in diameter,

Panoramic Video Synopsis in Timelapse

Panoramic Video Synopsis in Timelapse

Reducing spatio-temporal redundancy in dynamic video footage. Additionally, optical flow techniques are employed to generate a panoramic view, allowing for a comprehensive visualization of dynamic scenes.

Controllable Heliotrope Video

Controllable Heliotrope Video

The system follows the user's specified path, generating a continuous and controlled video sequence. A graph-based approach is used, where images are represented as nodes, and edges connect similar images based on appearance distance.