With China Mobile's investment in industrial AI infrastructure, the development of industrial visual inspection in intelligent manufacturing has entered a new stage. In the pharmaceutical industry, the quality inspection of capsules is crucial. WeLinkirt's AI AOI software system provides an effective solution for detecting multiple capsule defects.
User scenario: On the capsule production line of a leading pharmaceutical manufacturer, capsules are the main products, and the inspection objects are various filled capsules. In the capsule production process, strict inspections are required for the appearance, sealing performance and other aspects of the capsules to ensure that the product quality meets the high standards of the pharmaceutical industry.
Pain points: Under the traditional capsule inspection method, the manufacturer faces many difficulties. The miss-detection rate is relatively high, reaching about 3%, which causes some defective capsules to enter the market, posing certain safety risks. The false-alarm rate is also not to be underestimated, about 10%, which leads to a large number of qualified capsules being misjudged as defective products, increasing the workload of re-inspection and production costs. In addition, the model change time is long, and it takes about 30 minutes to change the product model each time, seriously affecting production efficiency. Combining with the current trend that China Mobile's investment in industrial AI infrastructure promotes the development of industrial visual inspection, the manufacturer urgently needs to introduce advanced inspection technology to improve the situation.
Technical principle
WeLinkirt's AI AOI software system uses an advanced feature recognition algorithm based on a visual foundation model. This algorithm can accurately identify the normal and abnormal features of capsules through deep learning and analysis of various features of capsules. The principle is that the visual foundation model has a deep understanding and memory of different types of capsule features after being trained with a large number of samples.
- The 0-code automatic programming function for a good product in 5 minutes enables the system to quickly adapt to new capsule models. By learning from one good-quality capsule, the system can automatically generate an inspection program in a short time without complex code writing, greatly shortening the model change time.
- The APDT positive-sample/few-sample learning (1-20 good samples) technology can accurately identify the normal features of capsules by using a small number of positive-sample capsules for learning and training. This is because this technology extracts and analyzes the features of positive samples to build an accurate feature model, which can also achieve precise detection of minor defects in capsules.
- The semantic false-alarm filtering function is based on semantic understanding technology to analyze and filter the detection results. It can identify false-alarm information caused by environmental factors or other interferences and filter it out, thus effectively reducing the false-alarm rate.
WeLinkirt's solution and product
Centered on the AI AOI software system, WeLinkirt provides a complete solution for the manufacturer. The system has strong self-training ability and can quickly adapt to the inspection needs of different capsule models. In the implementation process, first, use the APDT positive-sample/few-sample learning technology, and only 10 good-quality capsules are needed to complete the system training. Then, use the 0-code automatic programming function for a good product in 5 minutes to achieve rapid model change, and the model change time is shortened to 5 minutes. At the same time, the semantic false-alarm filtering function effectively reduces the false-alarm rate. In addition, the system also supports 100% local private deployment of SDK/API/Docker to ensure that data does not leave the factory, meeting the compliance requirements of the pharmaceutical industry. The supporting DaoAI 2D/3D AI AOI equipment can provide more comprehensive inspection data to assist the AI AOI software system in making more accurate judgments.
The AI AOI software system brings an efficient, accurate and safe solution for pharmaceutical capsule inspection.
Quantitative results: After introducing WeLinkirt's AI AOI software system, the manufacturer has achieved remarkable results. The miss-detection rate has been reduced from the original 3% to <0.6%, greatly improving product quality and reducing safety risks. The false-alarm rate has been reduced from 10% to <4%, reducing the workload of re-inspection and production costs. The model change time has been shortened from 30 minutes to 5 minutes, improving production efficiency and enabling the production line to more flexibly respond to the production needs of different products.
FAQ
How many good-quality capsules are needed for the training of the AI AOI software system?
The system uses APDT positive-sample/few-sample learning technology. Only 1-20 good-quality capsules are needed. In this case, 10 good-quality capsules can complete the system training and quickly build an accurate inspection model.
How much can the model change time of the system be shortened to?
Using the 0-code automatic programming function for a good product in 5 minutes, the model change time can be shortened from 30 minutes to 5 minutes, greatly improving the flexibility of the production line to handle different products.
How does the system reduce the false-alarm rate?
The system has a semantic false-alarm filtering function. Based on semantic understanding technology, it analyzes and filters the detection results, can identify and filter out false-alarm information caused by environmental factors, etc., and reduce the false-alarm rate.