As food processing becomes increasingly automated, a reliance on manual inspection methods can hold back process optimisation. In the following article, LMI Technologies explains how manufacturers can harness the power of 2D and 3D vision with edge AI.
Automating food processing presents significant inspection challenges due to irregular product shapes, low-contrast surfaces and the high throughput of organic materials. Traditional manual inspection cannot keep pace with modern production speeds, or the rigorous quality and food safety standards required for mass production.
To overcome these challenges, LMI Technologies delivers comprehensive, non-contact inspection solutions that combine metrology-grade 3D scanning and 2D imaging with advanced edge AI. By unifying 2D smart cameras, 3D smart sensors and on-device deep learning within the browser-based GoPxL® platform, manufacturers achieve intelligent inline inspection without external PCs or cloud dependency.
Maximising production yield and material utilisation
Advanced inline inspection uses high-resolution 3D measurement to capture complete shape data and compute precise volumetric measurements that traditional 2D vision cannot provide. Networked Gocator® laser profilers continuously scan raw proteins, fish, cheese blocks, and other irregular products to calculate accurate volumes and cross-sectional geometry.
This measurement data is transmitted instantly to downstream slicing and portioning equipment, enabling dynamic optimisation of cutting paths, improved slice consistency, higher product yield and reduced material waste.
Comprehensive defect and contaminant detection
Combining 2D intensity imaging with 3D height data creates a complete surface profile for reliable defect detection. While 3D smart sensors accurately measure shape and volume, 2D smart cameras capture subtle cracks, discoloration, foreign material and surface blemishes that height data alone may not detect.
The GoPxL® platform processes both inspection modalities within a single edge architecture. Using powerful tools such as GoPxL® Anomaly Detector, manufacturers can quickly train and deploy deep learning models directly at the edge, enabling real-time detection of defects in baked goods, packaging and other food products with minimal development effort. This raises the question of whether automated quality inspection is easier than you think.
End-to-end traceability and packaging integrity
Food safety depends on complete product traceability and packaging verification. Beyond inspecting package features such as seals and closures, smart vision systems automatically read, decode and validate barcodes, batch codes and expiry information.
Leveraging edge AI, Gocator® 2D Smart Cameras make real-time pass/fail decisions and immediately trigger rejection systems for mislabelled, damaged or non-compliant products.
Non-contact quality assurance at production speed
Non-contact inspection eliminates contamination risk while maintaining reliable performance in demanding food production environments, including wet and washdown applications. This reflects the wider role of robotics and automation in the food industry.
Powered by NVIDIA® Jetson Orin™ NX processors, LMI inspection systems process complex datasets at rates of up to 3 kHz, providing 100 per cent inline inspection that improves product quality, maximises material utilisation, reduces downtime and scales efficiently across production lines and facilities.
