Quick Summary
- ▸ CodeBranch built an AI-powered system for a sheet metal manufacturing plant that uses image recognition to catalog scrap waste by shape, dimensions, gauge and material — then applies genetic algorithms to match inventory to new projects and optimize cutting patterns to minimize future waste.
- Scrap pieces are automatically identified, measured, and cataloged using AI image recognition — no manual measurement required.
- Existing waste inventory is matched to new project requirements before ordering new material, reducing raw material costs.
Overview
A sheet metal manufacturing plant needed a way to stop discarding usable scrap. CodeBranch built a system that uses AI image recognition to photograph and catalog waste pieces automatically — extracting shape, dimensions, gauge, and material type without manual measurement. When a new production project starts, a genetic algorithm searches the waste inventory for pieces that match the project's dimensional requirements, enabling direct reuse before ordering new stock. When new sheets must be cut, the system suggests optimal cutting patterns to minimize the scrap generated. The result is a closed loop: waste is cataloged, matched, and reused — and new waste is minimized at the source. For the plant, this means lower material costs, less waste sent to disposal, and an inventory of available scrap that operators can search instead of guess about.
Industries
Services Provided
- Custom Software Development
- AI Development
- AI Image Recognition
Approach
CodeBranch assembled a team of 6 — two senior developers, two semi-senior developers, one QA expert, and one UI/UX designer. The AI image recognition pipeline was built in Python: scrap pieces are placed on a designated photography table, and the system processes each photograph to extract dimensional data automatically. Polygon geometry computation is handled by PolyK, which models the shapes of both scrap pieces and the parts required by new projects. The genetic algorithm engine (Genetics-JS) handles the matching and cutting optimization logic — searching the inventory for reusable pieces and calculating cut patterns that minimize waste from new sheets. The application and UI were built in JavaScript, giving plant operators a searchable inventory interface where they can see available scrap organized by shape, dimensions, gauge, and material, and review the system's matching and cutting recommendations before acting on them.
Results
- Scrap pieces are automatically identified, measured, and cataloged using AI image recognition — no manual measurement required.
- Existing waste inventory is matched to new project requirements before ordering new material, reducing raw material costs.
- Optimal cutting patterns are suggested for new metal sheets, minimizing waste generation from every production run.
- Plant operators work from a structured, searchable inventory of available scrap organized by shape, dimensions, gauge, and material type.
- The system closes the loop between waste generation and reuse — turning scrap from a disposal cost into a recoverable asset.
