You will get Computer Vision Defect Detection Algorithm Development
Project details
Production-Ready AI Defect Detection & Vision Measurement
Stop trying to run industrial quality control on Jupyter Notebook prototypes. I deliver production-grade computer vision inference pipelines built to run 24/7 on your factory floor, optimizing both deep learning and classical vision.
Whether you need semiconductor die inspection, PCB solder joint QA, or surface scratch detection, I handle the full development chain to integrate vision software directly with your automation hardware.
Core Capabilities:
Deep Learning Inspection: Custom model training (YOLOv8/YOLO26, UNet, ResNet, MaskCNN, ViT, etc.) based on your latency and accuracy targets.
Classic CV Measurement: High-precision industrial metrology, dimensional gauging, and edge-detection measurement tasks.
Edge Optimization: Model quantization and conversion to TensorRT, ONNX, or OpenVINO.
Factory Deployment: Ready-to-integrate C++/Python inference code optimized for NVIDIA Jetson, x86 IPCs, or Docker.
Stop trying to run industrial quality control on Jupyter Notebook prototypes. I deliver production-grade computer vision inference pipelines built to run 24/7 on your factory floor, optimizing both deep learning and classical vision.
Whether you need semiconductor die inspection, PCB solder joint QA, or surface scratch detection, I handle the full development chain to integrate vision software directly with your automation hardware.
Core Capabilities:
Deep Learning Inspection: Custom model training (YOLOv8/YOLO26, UNet, ResNet, MaskCNN, ViT, etc.) based on your latency and accuracy targets.
Classic CV Measurement: High-precision industrial metrology, dimensional gauging, and edge-detection measurement tasks.
Edge Optimization: Model quantization and conversion to TensorRT, ONNX, or OpenVINO.
Factory Deployment: Ready-to-integrate C++/Python inference code optimized for NVIDIA Jetson, x86 IPCs, or Docker.
What's included $9,800
These options are included with the project scope.
$9,800
- Delivery Time 42 days
- Number of Revisions 2
- Number of Model Variations 2
- Number of Scenarios 1
- Number of Graphs/Charts 4
- Model Validation/Testing
- Model Documentation
- Data Source Connectivity
- Source Code
Optional add-ons
You can add these on the next page.
Fast 28 Days Delivery
+$4,800
Additional Revision
+$3,900
Additional Model Variation
(+ 10 Days)
+$4,800
Additional Graph/Chart
(+ 7 Days)
+$980Frequently asked questions
About Jin
Full-Stack Machine Vision Automation Engineer for industry inspection
Piscataway, United States - 11:19 am local time
What I bring that most freelancers can't:
- I design the mechanical structure (SolidWorks / AutoCAD)
- I write the firmware (ESP32, Arduino, motor control)
- I build the host software (C++, Python, Qt real-time UI)
- I develop the machine vision pipeline (OpenCV, Halcon, lighting & optics design)
- I train and deploy deep learning models (YOLO, UNet, ResNet, ViT → TensorRT / ONNX)
That means when you hire me, you're not hiring a vision guy who needs a firmware guy who needs a mechanical guy. You're hiring the engineer who can design the entire system and make sure every layer talks to the next — because I've done it on 20+ production systems.
As a U.S. citizen but live 50/50 in the US and China, I can offer you not only solo engineering, but can mobilize a complete R&D team — structural, electrical, firmware, software, and algorithm engineers — delivering at less than half US-market rates. And with direct access to China's hardware supply chain, prototypes move from CAD to physical reality in days, not weeks.
Industries I've served:
Semiconductor · Optical Communication · Rare-Earth Magnets · Consumer Electronics · Food & Beverage · PCB Manufacturing · Pharmaceutical Packaging
Clients include listed companies such as 光迅科技 (Accelink, SZ:002281), 龙磁科技 (Longi Magnetics, SZ:300835), 源杰科技 (Yuanjie Semiconductor, SH:688498), and subsidiaries of China Electronics Technology Group.
I hold degrees from USTC (University of Science and Technology of China) and Rutgers University (New Jersey, USA), with prior research roles at a world-leading pharmaceutical company and at Zhejiang University.
Availability: Remote with occasional on-site visits. Fluent in English and Mandarin Chinese.
Steps for completing your project
After purchasing the project, send requirements so Jin can start the project.
Delivery time starts when Jin receives requirements from you.
Jin works on your project following the steps below.
Revisions may occur after the delivery date.
Requirements & Data Kickoff
Client provides labeled/unlabeled images of good/bad parts, clear defect definition criteria, cycle-time latency limits, and target deployment hardware specs.
Dataset & Annotation Strategy
We audit the data, establish strict labeling guidelines, and design a class-balancing plan to handle rare industrial defect categories effectively.