SensorX: The benchmark in bone-detection for a reason

Dual-energy X-ray verses photon-counting technology 

Good bone detection needs to differentiate between material types – regardless of the bone size and the thickness or density variation of the protein. 

Dual-energy X-ray systems – such as the SensorX - are the proven, trusted backbone of food inspection today. Dual-energy X-ray can distinguish bone from protein thickness variation with stable-physics-based measurements. Which is how it provides consistent, and reliable performance across production runs. This technology maintains reliable detection without the need for repeat calibrations, ensuring uninterrupted normal flow of production lines. Making it ideal for food processing where efficiency is key to profitability.  

While photon counting has advanced a long way since it was first studied in the early 1990s, the sensor technology is less stable than established dual-energy X-ray methods. Resulting in lack of details and ‘shifting’ mood of the sensor. In high-speed food processing lines this means there is a high rate of false-positives, and sensors that work differently from one day to the next. To maintain accuracy, photon-counting systems require regular calibrations, which can slow down production flow.

For inline poultry inspection, the time-tested, and finely tuned JBT Marel dual-energy X-ray technology delivers the clear images needed to accurately detect bones, with fewer calibration sensitivities at line speeds. This delivers optimized production flow, reliable detection accuracy, plus low false positives at speed.

“We asked several companies for demonstrations of their technology,” said Thada Chittham, General Manager Primary Processing, BRF Thailand. “JBT Marel's accuracy was significantly higher: about 98% opposed to 70%. So, the SensorX bone detection system is incredibly efficient. The other X-ray systems were far more general, whereas the SensorX was specifically designed to find bone.” An additional benefit of the SensorX is its capability to detect other contaminants such as metal, stone and glass.

Advanced image analysis based on physics not just AI pattern-matching

While AI-enhanced vision can add value in multiple areas of processing equipment, including detection, AI ‘black-box’ models trained in pattern recognition make generalizations when confronted with scenarios outside its training. This generalization, leads to uncertainties when introducing new types of defects or contaminate that was not in the training data set.

“It’s time-consuming and difficult to train an AI model for every eventuality. And it’s often the unspoken expectations that are the problems,” said Hrafnkell Eiríksson, Senior Software Engineer, Innovation R&D, JBT Marel. “When you ask a processor what they want to focus on of course they say small bones. But no one discusses what everyone feels is obvious, like a drumstick. Clearly, that is a ridiculous example, but it’s also true. AI models can only detect what they are specifically trained to see. If you add a new product line, a new cut style, the AI has to be trained for it.”

The algorithms behind SensorX image analysis rely on defined physics, separating soft tissue from bone or other hard contaminants to reveal low-contrast fragments, regardless of the cut style or product line. Making detection more reliable and flexible to future changes.

Throughput and oversight

Bone inspection isn’t just about seeing the contaminant, it’s about efficiently removing bone and recovering yield with minimal labor. SensorX has an integrated reject mechanism, operator rework stations, and automatic re-introduction for a second inspection, closing the loop so reworked product is verified before continuing downstream.

Why processors keep choosing the SensorX

For more than 20 years we have been continuously developing and fine-tuning the SensorX to ensure processors have an inspection system that results in fewer complaints, higher yields, and predictable, stable detection.

And we’ll continue to do so, incorporating new technology that improves performance and meets the needs and challenges of food processors.

SensorX sets a bone-inspection benchmark for a reason.