SANOVO TECHNOLOGY GROUP implemented a contactless, multi-camera deep-learning inspection system to improve defect detection accuracy, eliminate contamination risk from touch-based testing, and provide data insights for both egg grading centers and farms. Powered by Emergent’s 10GigE HR Series cameras and developed using the eSDK software development kit, the system captures 16 images per egg and classifies defects using deep learning algorithms at production speeds up to 250,000 eggs per hour.

Emergent spoke with Jorrit van Hof, Product & Business Manager, SANOVO TECHNOLOGY GROUP about the project and ongoing relationship with Emergent Vision Technologies.

SANOVO machines inspect hundreds of thousands of eggs per hour. Traditional crack detection systems used on egg grading machinery rely on physical contact with the egg to detect structural damage. Mr. van Hof described how these force-based systems introduce risks by potentially worsening cracks, creating false rejects, and increasing the possibility of cross-contamination. As egg producers demand higher traceability and data insights, SANOVO sought a non-contact inspection approach capable of high accuracy at full processing speed.

SANOVO developed deep-learning models capable of identifying cracks and shell defects, but the accuracy of these algorithms depended on the quality and consistency of the captured image. The camera system would need to deliver high resolution at high line speeds, with reliable multi-camera synchronization in an industrial environment.

Initial vendor investigations revealed that competing camera suppliers could not guarantee performance or provide analysis systems without significant cost and risk. Some suppliers proposed expensive test equipment with no refund option if performance requirements were not met. This created unacceptable engineering risk.

Emergent Vision Technologies offered a no-pay evaluation system and collaborated directly with SANOVO engineers, providing configuration recommendations and validating camera capabilities during early testing.

The fiber-connected HR Series 10GigE high-speed cameras demonstrated sufficient quality to support SANOVO’s deep-learning crack analysis. SANOVO’s deep learning algorithms were tested within the GUI-based eCapture Pro software. Once validated with the cameras, programming was transferred into the eSDK software development kit. Throughout the process, says van Hof, Emergent provided helpful advice and supported the company’s engineering team.

The Emergent ecosystem of high-speed 10GigE cameras working holistically with eCapture Pro and eSDK sped up the development process. Mr. van Hof says that “Emergent were confident that the whole thing would work and stood behind it. Rather than delivering isolated components, Emergent worked alongside SANOVO as a partner, helping tune camera parameters, synchronization timing and deep-learning capture conditions.”

The eSDK now provides the imaging stream used within the egg inspection machines sold to customers.

“Requirements were considered too high by multiple suppliers, but Emergent provided a solution and proved it during evaluation. The system has demonstrated reliable operation with high detection performance.”
— Jorrit van Hof, Product & Business Manager, SANOVO TECHNOLOGY GROUP

SANOVO implemented a 21-camera inspection structure on an 18-row grading configuration—capturing top, side, and bottom views of every egg as it rolls beneath the vision system. The 21 cameras capture 16 images per egg to create a complete surface profile. Synchronous camera triggering is controlled by instructions in the eSDK software.

All cameras stream high-resolution images over fiber optic cable to an industrial PC equipped with NVIDIA GPUs, making the design electrically interference-free, hygienically robust, and scalable for both grading centers and smaller farm-level applications. The technology ensures that powerful GPUs perform capture, processing, transfer and storage, so inspections are conducted properly and data is reliably logged.

The system classifies defects and transmits results to the machine controller for gate activation, so cracked or sub-classified eggs are rejected automatically without slowing the line. In high-capacity configurations, the system supports throughput up to approximately 255,000 eggs per hour while maintaining reliable classification performance.

Since this initial machine integration, SANOVO’s use of the HR 10GigE cameras has expanded to other inspection units with varying levels of throughput, and supporting camera, network, and computing hardware. This means SANOVO offers an appropriately sized machine for every type of egg producer. These include machines such as the GraderPro600 and OptiGrader 707. Existing machines can also be retrofitted with the new vision system.

Enhanced real-time data visibility supports more accurate supply contracts, improves brand quality, and increases revenue potential from premium-grade eggs.

SANOVO’s contactless inspection technology has improved grading reliability and reduced false rejection of high-quality eggs. Mr. van Hof described how false reject rates are lower than they have ever observed, and egg producers have earlier insights into flock health than before. One benefit for producers is that laying hens are not prematurely retired since the quality of their eggs can be observed with greater precision and reliability.

A new machine called the Farm Intelligence Unit can detect trends such as crack types and shell irregularities earlier in the production cycle. This versatile machine can be installed at a greater range of production facilities.

Emergent cameras are now installed on multiple SANOVO egg grading machines that enable various throughput levels, depending on the size of the facility. All are multi-camera systems requiring precise camera synchronization and high industrial reliability. The company’s engineering team are considering new applications for the second generation 10GigE camera series, EROS, which bring smaller sizes, lower power consumption and newer Sony sensors to Emergent’s complete ecosystem.