Welcome to Kunshan EEK Automation Equipment Co., Ltd.
Home/Materials /AI Visual Inspection Equipment for Genuine Leather

AI Visual Inspection Equipment for Genuine Leather

Genuine leather AI visual inspection equipment targets natural raw animal hides and finished genuine leather, performing inline grading by defect location and grade, identifying natural defects such as scars, insect holes, blood veins and loose grain, and outputting grade and OK/NG.

Materials Overview

Natural flaws are inherent, so the inspection goal is grading rather than zero defects

Quick answers

Genuine leather AI visual inspection equipment images natural animal hides and uses algorithms to identify natural defects such as scars, insect holes, blood vessels, loose grain and cracking, and grades them by the position and size of the defect. It outputs the usable grade and OK/NG results for raw hide grading and cut-piece optimization.

Genuine leather comes from animal hides, so natural injuries (scars, grub holes, blood veins, whip marks, growth wrinkles) are inherent, and "zero defect" like synthetic materials is impossible. The goal of inspection is not to reach zero, but Accurate Grading, Keeping Defects Outside Usable Areas.

The core of grading is "position + size". On the same hide, damage in a corner area may mean only a one-grade drop, while damage in the main material area may scrap the whole hide. The equipment must therefore not only judge defects but also mark them precisely by coordinate so that nesting software can avoid them.

Compared with synthetic leather, genuine leather has more complex texture, lower contrast, and greater batch variation. The same algorithm often has to be modeled separately for different leather types (cow, sheep, pig), which is why genuine leather inspection relies more on samples and measurement.

Applicable Type

Cowhide Top Grain Leather

Coarse texture with many defects; the focus is grading and nesting avoidance

Sheepskin

Thin and soft with fine texture; illumination and pixel equivalent are demanding

Pigskin

Pores are arranged in a triangular pattern; the feature is distinct but damage forms are varied

Corrected Grain / Lightly Corrected Grain Leather

Sanded and uniform in surface, with the focus on coating finish and show-through

Leather Crust Roll Stock

Continuous material feed, suitable for line-scan full-width grading

Common defect

Defect TypesTypical ManifestationsInspection Focus Points
ScarsLocal tissue healing marksGrading by position and area
Warble Holes / Insect HolesPinhole PenetrationTransmitted light or low-angle light forms a clear image
Blood Veins / Vein MarksSubcutaneous veins visible at the surfaceEasily confused with defects; the model must distinguish them
Loose grainLoose, collapsed grain layerLow-angle light highlights relief
CrackingDry cracking or finish crackingFocus on edges, corners and bends
Wrinkles / Fat WrinklesNatural WrinklesMust determine whether it exceeds the acceptance range
holeShort Shot or PenetrationBacklight Is the Most Direct
color difference Inconsistent shade within the same batchStable light source and color datum

inspection Workflow

Complete chain from loading to judgment

Visual inspection judgement chainThe complete chain from image acquisition to OK/NG judgement and PLC interlocking.image acquisitiontrigger captureTarget Positioningtemplate matchingfeature recognitionAlgorithm JudgementResult JudgementOK / NGindustrial communicationPLC interlockingRelease OKRejection / NG AlarmEvery step's result retains the image and judgement item, for traceability and review
Inspection Judgement Chain — Acquisition → Positioning → Recognition → Judgment → Communication interlocking; the entire chain runs locally on the machine.
  • 01 Loading and spreading of raw hide / cut pieces
  • 02 Flattening and de-wrinkling with tension control
  • 03 Multi-zone imaging covers the entire sheet
  • 04 Texture and defect segmentation
  • 05 Defect coordinate marking
  • 06 Grading by rule (position + size)
  • 07 Output Grade and OK/NG
  • 08 Coordinate-based feeding / rejection
  • 09 Archive images and grading results

AI Inspection Principles

AI Visual Inspection Principles

Genuine leather inspection follows a "grading logic" rather than a "presence/absence logic". After building a model of the normal grain texture, the algorithm marks out the coordinates of abnormal areas and then applies the customer's grade rules (where the defect falls, how large it is, which grade it counts as) to output the result. Success again depends on imaging: the relief of the hide, the reflection from pores, and the texture differences between leather types must all be suppressed.

For illumination, diffuse light is used to suppress pore reflection and reveal defect contours; low-angle light is used to expose the relief of loose grain and cracking. Stains on dark genuine leather (such as black split cowhide) and scratches on light-colored leather require separate lighting settings, so recipes are usually grouped by leather type and color.

The grading rules must come from the customer's acceptance specification, not be defined by the equipment manufacturer on its own. It is recommended to first perform grading measurements on a representative batch of leather and align the grading results with manual grading before going live.

Applicable Industry

Applicable equipment

Common Question

What Is the Difference Between Genuine Leather and Synthetic Leather Inspection?
Synthetic leather aims for "zero defects", while genuine leather aims for "accurate grading". Genuine leather has natural scars and flaws, and the inspection goal is to locate and grade them for nesting and avoidance, not to reject everything. The two differ in both illumination and algorithm model.
Can the equipment automatically grade leather?
It can output a grade suggestion based on coordinates and area, but the grading rules must be provided by the customer (where the defect falls, and how large it must be for each grade). It is recommended to first align machine grading with manual grading using samples.
Will natural markings such as blood veins and fat wrinkles be mistaken for defects?
Yes, and this is the main source of false calls. OK samples must be used to build a normal texture model, and these "normal irregularities" must be labeled as non-defects during training to keep over-rejection down. The final confirmation is the measured over-rejection rate.
Can defect positions be accurate enough for feed and nesting avoidance?
Yes. Defects are output as coordinates and can be integrated with nesting software to avoid the main material area. Positioning accuracy depends on camera resolution and calibration, and must be measured according to the actual area and the smallest avoidable size.
What is the inspection accuracy?
To be added. Accuracy is determined by the field of view and camera resolution: pixel scale = FOV ÷ pixel count. The minimum defect size must be defined first, then the imaging solution derived and measured with samples; it cannot be summarized as a single number.
Can It Be Integrated Into an Existing Production Line?
Yes. It communicates with the PLC via I/O, TCP, RS485, Modbus, S7, Profinet and others, outputting grade signals and OK/NG; the specific protocol is subject to on-site conditions.
What samples are needed to develop a solution?
Provide OK and NG samples in different leather types and colors, along with the customer's grading rules, defect acceptance criteria, loading pose and cycle time.

Submit sample testing

Send OK and NG samples of genuine leather from different hides (cow / sheep / pig) and in different colors, along with your grading rules; we will run grading measurements and give you an illumination plan plus recommendations on coordinate marking accuracy.

JPG / PNG supported, multiple files allowed
Each file must not exceed 20 MB
    JPG / PNG supported, multiple files allowed
    Each file must not exceed 20 MB
      JPG / PNG supported, multiple files allowed
      Each file must not exceed 20 MB
        A solution engineer will contact you within 1 business day after submission

        Need to assess imaging conditions, defect criteria and cycle time item by item? Go to the Full Requirement Assessment →

        Send Us Your Material and We Will Show You the Measured Results

        Provide several OK and NG samples of the material, and a solution engineer will run actual imaging and judgement tests on the equipment to give a workable inspection configuration recommendation, rather than reading the datasheet alone.

        Online support WhatsApp
        WeChat
        Scan the QR code to add us on WeChat
        Call us Back to top