Automated Quality Inspection for Canned Fish – How Vision Systems Support Food Manufacturers

As production line throughput continues to increase, manual inspection of cans in canned fish manufacturing is becoming an ever greater challenge. Scanway’s advanced vision systems inspect up to 300 cans per minute, providing 100% inspection coverage without slowing down production. Every can is analysed with maximum precision, regardless of line speed, making automated vision inspection significantly faster and more accurate than manual quality control.
Automated image analysis eliminates subjective operator assessment and ensures consistent inspection, as every can undergoes the same inspection procedure. As a result, canned fish manufacturers benefit from a stable, predictable and fully automated quality inspection process. In addition, through integration with an automatic reject system, every defective product can be removed from the production line immediately.

What Defects Can a Vision Inspection System Detect??

A modern quality inspection system should detect all defects that may compromise package integrity, product safety and overall product quality. Scanway’s vision system detects, among others:

• Seam defects: seam teeth, seam droop, dents, unbonded wrinkles, excess material protruding beyond the flange, seam cracks and cuts, cut-off flange, and seam overlap.
• Can geometry defects: dents and other deformations.
• Printing defects: incomplete printing, incorrect print for the production batch, and lid-to-can body mismatch.
• Label defects: missing, detached or misaligned labels.
• Easy-open key defects: missing or misaligned easy-open keys.

The CUMULUS vision system not only detects defects but also supports production process optimisation. All information on detected non-conformities is automatically collected and archived, enabling detailed quality analysis and the generation of comprehensive production statistics. By identifying product anomalies, the system can also help predict potential machine failures, supporting continuous production and reducing unplanned downtime.

How Does the Scanway Vision System Work?

The solution has been designed for seamless integration with existing canned fish production lines. It uses a set of infrared cameras combined with proprietary image analysis algorithms to inspect the quality of can seams.

Every can passing through the inspection station is analysed in real time and automatically classified into one of two quality categories:

• OK – product compliant with quality requirements.
• NOK – defect detected; product rejected.

When a defect is detected, the system can send a signal to the reject mechanism, which automatically removes the defective can from the production line. The entire inspection process is performed without affecting production line throughput.

The solution supports the most common can formats used in the canned fish industry, including Hansa, Club, and round cans, as well as a wide range of colours and graphic designs. In addition to real-time classification, the system archives images of detected defects together with process data, enabling detailed quality analysis and root cause identification.

Importantly, the solution has moved beyond the testing phase. Scanway’s vision systems are already deployed on canned fish production lines, supporting day-to-day quality inspection under real production conditions.

Case Study: Proven Performance Under Real Production Conditions

The system’s performance was validated on a live production line at a canned fish manufacturing facility. The validation was carried out primarily under normal production conditions at full line speed, complemented by controlled tests using pre-prepared cans containing seam defects. In total, the system inspected more than 100,000 cans, covering all packaging formats used by the manufacturer, including Hansa, Club and round cans, as well as a wide range of lithography, colour and construction variants. This enabled comprehensive verification of the system’s performance across diverse production conditions and the manufacturer’s entire product portfolio.

Scanway’s vision system automatically classifies every can into one of two quality categories: OK (product compliant with quality requirements) and NOK (seam defect detected), in accordance with the manufacturer’s quality specifications. Performance analysis demonstrated a 99.8% classification accuracy (Accuracy), meaning that virtually all inspected products were classified correctly. The system also achieved an exceptional 99.96% recall, confirming its ability to detect virtually every can with a seam defect while minimising the risk of defective products progressing to subsequent stages of the production process.

Another key performance indicator is the system’s 91.8% precision, demonstrating a high level of reliability in identifying actual defects. At the same time, the False Positive rate—products meeting quality requirements but incorrectly classified as defective—was only 0.2% of total production, helping minimise material losses while maintaining high production efficiency. The False Negative rate, representing cans with seam defects that were not detected by the system, was just 0.001%, confirming the solution’s exceptionally high reliability.

During validation, the system also analysed 2,413 real seam defect cases, which were used to evaluate the performance of the inspection algorithms. The solution’s reliability was further confirmed by the absence of customer complaints related to the inspected production batches, while comparative inspections carried out by operators identified no additional classification errors. In addition, the system collects and archives inspection data, providing manufacturers with valuable insights to support quality monitoring, process optimisation and production decision-making.

A modern vision system does more than simply classify products as OK or NOK. The captured images and process data provide a valuable source of production quality information, enabling manufacturers to analyse recurring defects, identify their root causes more quickly, and monitor process stability over time.

By combining automated quality inspection with data analysis, canned fish manufacturers gain more than highly effective defect detection. They also benefit from a powerful tool for optimising production processes, maintaining consistent process performance, and ensuring the highest quality standards.