陈晋，英国伦敦开发人员
Chen is available for hire
Hire Chen

Chen Jin

Verified Expert  in Engineering

Computer Vision Developer

Location
London, United Kingdom
Toptal Member Since
June 30, 2022

Jin has multidisciplinary experience developing deep learning and computer vision methods for medical, auto-piloting, and geospatial image sectors, with works published at ICLR, NeurIPS, and MICCAI. He developed data-and-computation-efficient machine learning methods for budgeted computer vision tasks such as segmentation, detection, classification, super-resolution, GAN, and 3D reconstruction. Jin also excels at multimodal self-supervised representation learning, image generation, and mapping.

Signal ProcessingComputer VisionMachine LearningArtificial Intelligence (AI)Deep LearningAlgorithmsBatchPythonJupyterLinuxPyTorchCloudMATLABTensorFlowCChemical EngineeringGPT-3Prompt EngineeringImageJ

Portfolio

University College London
Python, PyTorch, MATLAB, R, TensorFlow, Jupyter, Keras, Linux, Cloud, Cluster...
University College London
Python, PyTorch, MATLAB, R, TensorFlow, Jupyter, Keras, Linux, Cloud, Cluster...
University College London
Python, PyTorch, R, MATLAB，算法，人工智能(AI)...

Experience

Machine Learning - 5 yearsDeep Learning - 5 yearsArtificial Intelligence (AI) - 5 yearsComputer Vision - 5 yearsPython - 5 yearsPyTorch - 4 years通用人工智能(AGI) - 2年

Availability

Part-time

Preferred Environment

Python, PyTorch

The most amazing...

...我开发的方法是用于分割的自适应降采样, 它提高了10%的精度，同时比SoTA节省了90%的计算.

Work Experience

多模态学习与图像生成研究员

2021 - PRESENT
University College London
  • Developed a multimodal-multiscale MRI-histology-text self-supervised representation learning method and bypassed the need to use expensive and destructive histology images for the first time.
  • 创建了一个跨模态尺度的mri -组织学-文本图像生成和映射(高档1),000 times) method for explainable AI.
  • Developed an interactive web UI based on a Gradio API to showcase registered cross-modality image patches.
  • Co-developed a disentangling inter-reader variability in segmentation method and published it at MICCAI/NeurIPS 2020.
Technologies: Python, PyTorch, MATLAB, R, TensorFlow, Jupyter, Keras, Linux, Cloud, Cluster, ImageJ, Batch, Computer Vision, Machine Learning, Artificial Intelligence (AI), Image Processing, Medical Imaging, 3D Image Processing, Deep Learning, Docker, GPT, Natural Language Processing (NLP), Generative Pre-trained Transformers (GPT), Algorithms, Data Science, Computer Vision Algorithms, User Interface (UI), Web UI, Gradio, Convolutional Neural Networks (CNN), Health, Signal Processing, Models

Researcher

2019 - 2022
University College London
  • Developed an attention-based deformable downsampling method for an end-to-end optimization of low-cost segmentation, 提高了10%的精度，同时比SoTA节省了90%的计算, 并在ICLR 2022上发表了一篇论文.
  • Created an attention-based learnable "patch loader" to sample the best size-resolution trade-off configuration at each location for optimal segmentation accuracy, achieved SoTA, 并在MICCAI 2020上发表了一篇论文.
  • 参与了三次超高分辨率图像分割挑战, 开发和部署各种SoTA方法, 进入了所有挑战的前10%.
  • 对高分辨率十亿像素图像进行高效分割.
Technologies: Python, PyTorch, MATLAB, R, TensorFlow, Jupyter, Keras, Linux, Cloud, Cluster, ImageJ, Batch, Computer Vision, Machine Learning, Artificial Intelligence (AI), Image Processing, Medical Imaging, 3D Image Processing, Deep Learning, Docker, Natural Language Processing (NLP), GPT, Generative Pre-trained Transformers (GPT), Team Leadership, Algorithms, Data Science, Computer Vision Algorithms, Convolutional Neural Networks (CNN), Health, Signal Processing, Models

Researcher

2020 - 2021
University College London
  • 共同开发了在MIDL 2022和MICCAI 2022上发表的半监督伪标记方法.
  • 利用深度学习为癌症成像做出了贡献. 开发了一种用于廉价MRI单模态推理的自监督表示学习方法.
  • Hosted the MedICSS 2021 Summer School project and taught a one-week deep learning course for segmentation, including theory and coding, led a team of seven researchers, 在14个团队中获得项目前三名.
Technologies: Python, PyTorch, R, MATLAB，算法，人工智能(AI), Deep Learning, Data Science, Computer Vision Algorithms, Convolutional Neural Networks (CNN), Health, Signal Processing, Models

Developer and Supervisor

2019 - 2021
University College London
  • Supervised one PhD student working on deep multiple instance learning (MIL) for histological image classification, 谁的最新作品提交给了期刊.
  • Supervised and supported two MSc students performing CNN-based segmentation and detection on histology images. One obtained distinctions.
  • 开发了一种多尺度循环超分辨方法. Supervised and provided technical support to two MSc students working on super-resolution with CNN and GAN-based methods. Both obtained distinctions.
Technologies: Python, PyTorch, Jupyter, Team Leadership, Algorithms, Artificial Intelligence (AI), Deep Learning, Data Science, Object Detection, Computer Vision Algorithms, Object Tracking, Convolutional Neural Networks (CNN), Health, Signal Processing, Models

Research Associate

2017 - 2018
Heriot-Watt University
  • Acted as the research associate working on computational geoscience and 3D modeling of porous media.
  • 开发了一种基于滤波器组的弱监督迭代卷积网络, AdaBoost, and auto-context, 提高细线状裂缝的分段连通性. 它在2016年和2018年的会议上提出.
  • Contributed to the CT and SEM images registration and multiscale data fusion 并利用多点统计方法指导三维孔隙结构重建 conditional to neighboring "patch."
  • 开发了一种基于随机森林的基于斑块的岩石模式自动分类程序.
  • Completed the modeling and numerical simulation of macroscale reservoir multiphase flow and presented it to the industry.
Technologies: MATLAB, C, C++, JavaScript, Batch, ImageJ, ParaView, Cinema 4D, Computer Vision, Machine Learning, Artificial Intelligence (AI), Image Processing, 3D Image Processing, Algorithms, Data Science, Computer Vision Algorithms, Geology, Signal Processing, Models

Engineering Intern

2013 - 2014
Canadian Natural Resources Limited (CNRL)
  • 基于Eclipse建立油藏流体流动模拟模型.
  • 通过模拟研究了各种地质层状构造的影响.
  • Performed an analysis of a zonal allocation and potential reservoir in the south Ninian area and presented it to a development team that helped with drilling decisions worth millions.
Technologies: Eclipse, Models

超高分辨率图像的下采样分割

http://lxasqjc.github.io/learn-downsample.github.io/
We introduced an approach for learning to downscale high-resolution images for segmentation tasks. The main motivation was to adapt the sampling budget to the difficulty of segmented pixels or regions. We showed that learning the spatially varying downsampling strategy jointly with segmentation offered advantages in segmenting large images with a limited computational budget.

作为主要作者，我全权负责开发工作. This work was published at ICLR 2022.

百万像素组织学图像的注视点分割

http://chenjin.netlify.app/publication/foveation/
Segmenting ultra-high resolution images often needs empirical decisions on the trade-off patch configuration between field-of-view (FoV) — spatial coverage — and the image resolution. We introduced the foveation module, a jointly learnable data loader which, for a given ultra-high resolution image, adaptively chooses the appropriate configuration (FoV or resolution trade-off) of the input patch to feed to the downstream segmentation model at each spatial location of the image.

这项工作发表在MICCAI 2020上，我是监督开发的主要作者.

超高分辨率图像的注视点分割

http://chenjin.netlify.app/publication/foveation_ultrahigh/
在MICCAI 2020之前的创新工作之上, in this extended version, we further introduced the more computational efficient hard-gated categorical sampling of FoV-resolution patch configurations at each location and provided two differentiable solutions. 我们演示了三个远景基准的一般适用性, including Cityscapes autopilot images, DeepGlobe aerial images, and Gleason2019 histopathology images.

作为监督整个开发过程的主要作者，我在Arxiv上发表了这项工作.

多模态，多尺度，多任务表示学习，图像生成和映射

ROLE AND RESPONSIBILITIES
•开发了一种自我监督的MR组织学文本表示学习方法, 目标是廉价的MRI单模态推断.
•创建了跨模态尺度的MRI组织学图像生成和映射(升级1),000 times) for explainable AI.
• Developed an interactive web UI based on a Gradio API to showcase registered cross-modality image patches.
• Disentangled inter-reader variability in segmentation and co-authored the MICCAI/NeurIPS 2020 publication.

癌症成像分析的分割、分类、检测、超分辨率和GAN

ROLE AND RESPONSIBILITIES

• Supervised a PhD student on deep multiple instance learning (MIL) for large image classification and submission to journals.
• Supervised two MSc projects performing CNN-based segmentation and detection of histology images.
•开发和监督两个基于CNN和gan方法的超分辨率硕士项目.

多尺度模态图像融合、重建和流体模拟

ROLE AND RESPONSIBILITIES

•开发了基于滤波器组的弱监督迭代卷积网络, AdaBoost, 自动上下文，提高细线状裂缝的分段连通性. 它在2016年的会议上展出，并在2018年以海报的形式展示.
• Conducted CT and SEM image registration, multiscale data fusion, 并利用多点统计方法指导三维孔隙结构重建, conditional to neighboring "patch."
• 开发了一种基于随机森林的基于斑块的岩石模式自动分类程序.
• Conducted modeling and numerical simulation of a macroscale reservoir multiphase flow presented to the industry.

医学图像分析挑战- Gleason19

http://gleason2019.grand-challenge.org/
ROLE AND RESPONSIBILITIES

•编写R脚本，将六个专家的像素注释合并为一个.
• Developed and trained deep learning models for the segmentation and classification of prostate cancer Gleason grading on ultra-high-resolution histopathological tissue microarray (TMA) images.
• Developed evaluation codes to run on the test dataset as per the task request and submitted them to the competition. Entered the top ten on the leaderboard.

医学图像分析挑战赛- DigestPath2019

http://digestpath2019.grand-challenge.org/
ROLE AND RESPONSIBILITY

• Developed and trained a segmentation model for ultra-high-resolution histology image segmentation and classification.
• Handled the preprocessing of ultra-high-resolution histology images into paired patches suitable for a deep learning model.
•开发评估代码，并将其打包在Docker容器中提交.

地理空间图像分析挑战-远程人工智能

ROLE AND RESPONSIBILITY

•包括遥感图像数据清洗和预处理.
•部署各种细分模型，以平衡性能效率, such as U-Net, MobileNet, and HR-Net.
•实现各种优化技巧，以提高性能, such as augmentation and regularization.
•根据组织者的测试要求开发测试代码和测试包.
•进入决赛(前10%).

Reservoir Simulation Based on Eclipse

对两相(水/油)油藏进行了模拟. The models have regular and irregular shapes with wells at opposite corners to simulate production in a quarter-five-spot pattern. Providing reservoir properties, history matching, 用Eclipse模拟水/油驱替.

Languages

Python, Batch, C, R, C++, JavaScript

Libraries/APIs

PyTorch, TensorFlow, Keras

Tools

木星，集群，ImageJ, TensorBoard, MATLAB, ParaView, Cinema 4D

Platforms

Linux, Docker, Eclipse

Other

Computer Vision, Machine Learning, Artificial Intelligence (AI), Image Processing, Deep Learning, Algorithms, Computer Vision Algorithms, Convolutional Neural Networks (CNN), Health, Signal Processing, Models, Cloud, 3D Image Processing, Artificial General Intelligence (AGI), Team Leadership, Object Detection, Object Tracking, Geology, Simulations, Medical Imaging, Satellite Images, Data Representation, Natural Language Processing (NLP), User Interface (UI), Web UI, Gradio, Engineering, Formation Evaluation, Economics, Test Analysis, Calculus, Linear Algebra, Chemical Engineering, Hydrology, Fluid Dynamics, Civil Engineering, Oil & 采购产品天然气，管道，存储，GPT，生成预训练变压器(GPT)，提示工程

Paradigms

Data Science

Industry Expertise

Petroleum Engineering

2013 - 2017

石油工程-计算地球科学博士

赫瑞瓦特大学-英国苏格兰爱丁堡

2011 - 2012

Master's Degree in Petroleum Engineering

赫瑞瓦特大学-英国苏格兰爱丁堡

2007 - 2011

石油工程学士学位

中国石油大学(青岛

AUGUST 2018 - PRESENT

CS231n: Deep Learning for Computer Vision

Stanford University

JUNE 2018 - PRESENT

CS221人工智能:原理与技术

Stanford University

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

在与Toptal领域专家的电话中讨论您的需求并细化您的范围.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

与你选择的人才一起工作，试用最多两周. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring