Skip to content
E3TAM

Multiclass weld-bead morphology classification

Solution concept

Regular, wavy and interrupted weld morphologies can have the same average width. Process classification needs a separate assessment of the bead image structure.

Artificial intelligenceresearch-backed conceptraw image / analysis separationclass confidence gate
Information delivered

Add the visual morphology class and confidence gate to the coupon record; transfer a class mismatch for review. Visual class does not establish weld strength.

Multiclass weld-bead morphology classificationILLUSTRATIVE SIMULATION
Multiclass weld-bead morphology classification
  1. 01 Recipe
  2. 02 Raw image
  3. 03 Class scores
  4. 04 Confidence gate
  5. 05 Route
—

Load the class expected by the order.

Top class score
—
Top-two class gap
—
Hold for review
—

Scenes are illustrative simulations; they are not customer references or performance guarantees.

Solution approach

The controlled image ROI is classified; the raw image and C1/C2/C3 scores appear separately. The order class and confidence threshold are evaluated together. Index the coupon and link bead texture to C1/C2/C3 bars. An ambiguous morphology is shown as UNKNOWN to prevent confident-looking acceptance.

Process steps

  1. 01

    Recipe

    Load the class expected by the order.

  2. 02

    Raw image

    Trigger the specimen under fixed lighting and link the image to its ROI.

  3. 03

    Class scores

    Reveal the three illustrative class scores as comparative bars.

  4. 04

    Confidence gate

    A low score becomes UNKNOWN; a confident wrong class is also not accepted.

  5. 05

    Route

    Release the accepted example and transfer the others to the review nest.

Vision system design

Let us assess your application together

We test your sample in our laboratory and report on resolution, optics and lighting selection.

+90 212 275 2280 – 81
Esentepe / İstanbul · İkitelli OSB Laboratory
Request Feasibility