The leather manufacturing sector relies on manual classification systems that are subjective, inconsistent, and unable to detect microscopic surface variations. Quality is highly variable. Defects are difficult to see. Grading depends on individual experience and lighting conditions.
The result: inconsistent quality control, material waste, and production delays. Perseus is exploring whether spectral imaging technology can provide an objective, scalable solution.
Hyperspectral cameras capture hundreds of wavelength bands simultaneously — far beyond what the human eye perceives. Each material surface generates a unique spectral signature, a fingerprint of its molecular composition, texture, and structure.
Hyperspectral camera scans the leather surface across hundreds of wavelength bands simultaneously.
Each scan generates a unique spectral signature encoding the material's molecular composition and surface structure.
Machine learning models compare spectral signatures against a reference database to detect anomalies and classify quality.
Results are stored in a spectral material database, building an intelligent reference library over time.
Automated identification of surface defects, scratches, grain irregularities and structural anomalies that are invisible under standard lighting conditions.
Objective, consistent quality grading based on spectral data rather than subjective visual assessment. Eliminates inter-operator variability.
Integration into production lines for real-time quality monitoring. AI models flag non-conforming materials before they enter the manufacturing process.
A growing database of spectral fingerprints for different leather types, tanneries, and treatments. A long-term infrastructure for intelligent material management.