Fastest
Eureka on the web
When you want the answer in the next five minutes.
The agent works the prompt against patents and technical literature, citing every source.
Run your analysis now →https://www.patsnap.com/resources/blog/rd-blog/surface-defect-detection-by-machine-vision-patent-landscape/ · Patsnap · data cut-off 2026-07-31 · downloaded from the live page
Patent Landscape · Machine Vision InspectionSurface Defect Detection by Machine Vision Patents
- Filings doubled from 2021 to 2024, climbing from 15 to 30 published records a year before the usual 18-month publication lag starts to understate the newest filings.
- The leader holds 13 records, but the top 5 assignees combined account for only 19.7% of the 157 records in scope — concentration is modest and a long tail of single- and few-filing entrants remains active.
- Image processing and AI classes dominate the claims, with G06T and G06N each touching roughly half of all records, while dimensional and structural testing classes like G01B and G01M stay under 9%.
157
Published Records
20%
Top-5 Share of All Records
+100%
Filing Growth 2021→2024
US
Leading Jurisdiction
Filing growth compares 2021 (15 records) with 2024 (30) — a three-year span. 2024 is the most recent year we treat as complete: publication lags filing by roughly 18 months, so 2025 onwards are still filling in and any growth rate that ends there would understate the field. Top-5 share is the combined record count of the five largest assignees divided by all 157 records in scope (CR5), not by the ranked leaders only.
Published byPatsnap Research··7 min readSourced from Patsnap Eureka
Overview
What this landscape covers
Surface defect detection by machine vision sits at the intersection of optical inspection hardware and learned classification models. The search underlying this page combines core inspection terms — surface defect detection, machine vision inspection, automated optical inspection — with the technical qualifiers that separate production-grade systems from lab demonstrations: illumination geometry, false rejection rate, small sample learning, anomaly detection method, data augmentation and inline cycle time.
That combination filters out generic computer-vision filings and keeps records where inspection throughput, lighting setup or training-data scarcity is explicitly claimed — the constraints that actually decide whether a detection system survives a production line.
Records in scope, 2015–2026
Source: Patsnap Eureka. Assignee ranking and totals. Derived from a Patsnap search on Surface Defect Detection by Machine Vision covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Let an AI agent run this analysis on your own technology
Pick a task. Every answer cites the patents behind it.
10,000 free credits to start
The Data
Filing trends and technology composition
157 published records, spanning receiving offices from the United States to Taiwan, chart how surface defect detection claims have shifted from optical hardware toward learned models.
Filing trend: acceleration through 2024
Annual filings rose from a single record in 2017 to a peak of 30 in 2024, with the 2021-to-2024 span alone showing filings double from 15 to 30. Years after 2024 will read lower in this chart simply because publication has not caught up with filing — treat 2025 and 2026 as incomplete, not as a slowdown.
Technology composition by IPC subclass
Image data processing (G06T, 49.7% of the 157 records) and AI-based computing (G06N, 44.6%) are the two largest classes, followed by image and video recognition (G06V, 35.0%). Material analysis (G01N, 24.8%) and dimensional or structural measurement classes (G01B, G01M, under 9% each) are present but comparatively thin, which is where hardware-adjacent claims have more room to move.
Shares are the percentage of the 157 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Source: Patsnap Eureka. Filing trend and technology composition. Derived from a Patsnap search on Surface Defect Detection by Machine Vision covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.
Go deeper on Surface Defect Detection by Machine Vision with Eureka
This page is one run against one query. Ask Eureka your own question about surface defect detection by machine vision and every answer comes back with the patent numbers behind it.
Try EurekaKey Patents
Representative filing and most-cited prior art
Representative Record
Surface defect detection method based on augmentation training of deep neural network (AU2021106389A4, Fudan University, 2021-11-04)
FUDAN UNIVERSITY
The invention belongs to the technical field of surface defect detection, in particular to a surface defect detection method based on deep neural network with augmentation training. The surface defect data set is augmented to train a deep neural network in two steps of static augmentation training and dynamic augmentation training. In the first training step, the data augmentation only with rollover and small translation of the original data is performed for network training. While in the second step, other augmentation operations are included, such as translation, rotation, scaling, adding noise, brightness adjustment, image filling, etc.Abstract text as published; claim scope should be verified against the full family before drafting around it.
View full recordMost-cited records in this landscape
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | US20200175352A1 | Structure defect detection using machine learning algorithms | 314 |
| 2 | US6273338B1 | Low cost color-programmable focusing ring light | 250 |
| 3 | US5095204A | Machine vision inspection system and method for transparent containers | 205 |
| 4 | WO2018165753A1 | Structure defect detection using machine learning algorithms | 149 |
| 5 | US20210370993A1 | Computer vision based real-time pixel-level railroad track components detection system | 145 |
| 6 | US5936353A | High-density solid-state lighting array for machine vision applications | 137 |
| 7 | US20040008877A1 | Systems and methods for inspection of ophthalmic lenses | 71 |
| 8 | US20160321594A1 | Correlation between manufacturing segment and end- user device performance | 63 |
| 9 | US20230125477A1 | Defect detection using one or more neural networks | 55 |
| 10 | CN114092389A | 一种基于小样本学习的玻璃面板表面缺陷检测方法 | 50 |
Citation counts favour older, widely-referenced records inside this corpus — read them as a signal of influence on later filings, not as a ranking of current relevance.
Patent titles are shown in the language they were filed in, not translated, so that each record stays verifiable against the original filing — a translated title will not match in Eureka or in any national register. Each row carries its publication number; clicking a row searches Eureka by that number.
Source: Patsnap Eureka. Citation counts and representative records. Derived from a Patsnap search on Surface Defect Detection by Machine Vision covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Run it yourself
Put your own technology through the same analysis
Where to run it
For builders
MCP server & REST API
When it has to run inside your own pipeline.
Patent search, landscape analysis and assignee resolution as MCP tools. Drop them into any agent framework, or call REST directly.
Browse MCP servers →Insights
What the numbers mean for filing strategy
Three patterns stand out once filing volume, technology class and geography are read together.
Growth
+100%
2021 → 2024 filings
Momentum is real, not noise
Filings went from 15 in 2021 to 30 in 2024, a doubling inside three years and the fastest sustained rise in this dataset's history. That growth sits almost entirely in image-processing and AI classes rather than in optical hardware, meaning the competitive question has shifted from lighting rigs to model architecture and training data.
2025–2026 figures are still filling in due to publication lag.
Concentration
19.7%
top 5 share of 157 records
No single assignee controls the field
The leading assignee holds 13 records and the top 5 combined reach 19.7% of the 157 records in scope — meaningful but far from dominant. A long tail of single-filing universities, component makers and inspection vendors keeps the field open to new entrants who claim a specific defect type or process step.
Top 10 combined reach 32.5% of all records.
Technology mix
49.7%
records touching G06T
Claims cluster on image processing, not measurement
Image data processing (G06T) and AI-based computing (G06N) each appear in close to half of all records, while dimensional measurement (G01B) and structural testing (G01M) sit under 9%. That gap suggests inspection systems are being claimed primarily as software pipelines, leaving measurement-hardware integration comparatively under-described.
Shares sum above 100% because records carry multiple IPC classes.
Eureka AI Agent
Looking for what nobody has claimed yet?
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to surface defect detection by machine vision, with the prior art for and against each one.
Source: Patsnap Eureka. Co-assignee relationships and derived observations. Derived from a Patsnap search on Surface Defect Detection by Machine Vision covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Players
Who is filing, and where the claim space is thin
The ranked list spans automotive component makers, optical hardware vendors, semiconductor equipment firms and university labs — no single sector owns the field.
Leader
13 records
top-ranked assignee
A modest lead, not a monopoly
The top-ranked assignee holds 13 of the 157 records in scope, roughly four times the fifth-place count of 4. That gap between first and fifth place is where a challenger with a narrow, well-drafted claim can still carve out defensible space.
Fifth and tenth place both sit at 4 records.
Long tail
100 companies
ranked assignees
A wide, fragmented field
The ranking returned by the data endpoint covers 100 companies, most holding only one or two records. That fragmentation across universities (South Carolina, Manitoba), equipment vendors (Presscue Tech, Bystronic Laser, Applied Materials) and component suppliers (Bosch, BYK Chemie) signals a technology still being claimed from many directions at once.
Top 10 combined hold 32.5% of all 157 records.
Geography
44 filings
United States receiving office
US and China lead filing venues
The United States receives the most filings at 44, with China close behind at 35 and India at 22. Europe and the WIPO/PCT route each carry 17, showing applicants are pursuing broad multi-jurisdiction coverage rather than concentrating in a single office.
Taiwan trails at 5 filings.
Recent-year filing momentum
| Assignee | Recent year | YoY |
|---|---|---|
| City University of Hong Kong | 1 | — |
| Robert Bosch GmbH (Germany) | 0 | — |
| Ocular Sciences, Inc. | 0 | — |
| University of Manitoba | 0 | — |
| Inventec Corporation | 0 | — |
| Wistron Corporation | 0 | — |
| Bystronic Laser AG | 0 | — |
| BYK Chemie GmbH | 0 | — |
Source: Patsnap Eureka. Assignee-level momentum. Derived from a Patsnap search on Surface Defect Detection by Machine Vision covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
What's Next
Where to take this analysis
The dataset points to specific next steps depending on whether the goal is freedom-to-operate, portfolio building or competitive tracking.
Check freedom-to-operate against the densest classes
With G06T and G06N each touching roughly half of all 157 records, any new filing on image-based defect classification should be checked against this dense claim space before drafting.
Run a freedom-to-operate check in Eureka →Draft around the thinner measurement classes
G01B and G01M sit under 9% of records each, leaving room for claims that fuse dimensional measurement with learned vision models rather than treating them as separate systems.
Explore white space in Eureka →Track the fragmented leader board
With the top assignee at only 13 records out of 157, watching new entrants as they cross from one or two filings into a sustained program is a better early-warning signal than tracking the incumbent leader alone.
Set up assignee monitoring in Eureka →Source: Patsnap Eureka. Forward-looking reading of the same dataset. Derived from a Patsnap search on Surface Defect Detection by Machine Vision covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
FAQ
Questions practitioners ask about this landscape
This landscape maps 157 published records dated between 2015 and 2026 that combine surface defect detection, machine vision inspection or automated optical inspection language with technical qualifiers such as illumination geometry, false rejection rate or data augmentation. That is a filtered, purpose-built count rather than every patent that mentions machine vision, so it should not be compared directly to broader computer-vision patent counts. The filtering is what makes the 157 figure useful for competitive analysis rather than a raw keyword hit count.
The ranked leader holds 13 of the 157 records in scope, with the top 5 assignees combined reaching 19.7% of all records — a real lead but not a dominant one. The field includes automotive component makers, optical hardware vendors, semiconductor equipment firms and university labs, and the ranking runs to 100 companies, most holding only a handful of records each. That spread means tracking a single incumbent misses most of the competitive activity in this space.
Filings grew from 15 records in 2021 to 30 in 2024, a documented doubling over three years and the clearest growth signal in this dataset. Counts for 2025 and 2026 appear lower in the raw trend, but that reflects the roughly 18-month lag between filing and publication, not an actual slowdown. Any read of the most recent one to two years should be treated as provisional until later publications catch up.
Image data processing (G06T) and AI-based computing (G06N) are the two largest classes, each touching close to half of the 157 records, followed by image and video recognition (G06V) at 35.0%. Dimensional measurement (G01B) and structural balance testing (G01M) each sit under 9% of records, making them comparatively under-claimed relative to the software-heavy core of the field. That gap is a reasonable starting point for identifying where a hardware-integration claim might still find open space.
AU2021106389A4, filed by Fudan University, describes a surface defect detection method using a two-step augmentation training process for a deep neural network: static augmentation with rollover and small translation first, then dynamic augmentation adding rotation, scaling, noise and brightness adjustment. It is a representative example of how data-augmentation claims are being structured in this field rather than a dominant blocking patent. Anyone building a small-sample or augmentation-based training pipeline for defect detection should review its specific claim language before assuming clear space.
Answers are grounded in the same dataset. Derived from a Patsnap search on Surface Defect Detection by Machine Vision covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Research Surface Defect Detection by Machine Vision in depth with Eureka
Go past this page: query the whole surface defect detection by machine vision corpus yourself, in your own scope.
Every answer comes back with patent numbers you can open.
Disclaimer. This page is generated from Patsnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. Patsnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.
Machine translation. Assignee and organisation names originally recorded in Chinese, Japanese or Korean have been rendered into English by an AI translation step so that the tables stay readable. These renderings are best-effort and may not match a company's registered English name; the original name is what the underlying patent record carries, and it is what any Eureka query launched from this page uses.