Metal Additive Manufacturing Porosity Images from 81 Steel Cubes

81 additively manufactured cubes of a high manganese steel provide the source for this image dataset. Optical microscopy images are cropped to 600x900 slices. Marc Ackermann created this dataset, and associated code for processing and training ML models is available in a linked GitHub repository.

Use Cases

• Train object detection models to identify porosity defects based on optical microscopy images.
• Develop image segmentation models to quantify porosity area and distribution in metal samples.
• Benchmark machine learning models for quality inspection in additive manufacturing workflows.

Strengths

• Images originate from 81 distinct physical samples, providing a basis for material variation.
• Images are pre-processed into a consistent 600x900 pixel format.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

Marc Ackermann via paperswithcode

Collection Method

Optical microscopy images cropped from additively manufactured cubes.