Ai Visual Inspection System For Biodegradable Bagasse Round Plate

Inspection standards

Application of camera

1. Front area detection: The reverse side of the incoming material on the tray is upward, and it is suspended in the air by the negative pressure belt. A set of industrial cameras is arranged directly under the tray, and the use of visual light source is used to collect images of the front area for processing.
2. Reverse area detection: After the front area is detected, the pallet falls on the rear belt smoothly, and the reverse side is still facing up. A set of industrial cameras is arranged directly above the pallet, and with the use of visual light source, the image of the reverse area is collected for processing.
3. Bevel area detection: The reverse side of the pallet is facing upwards, and 4 sets of industrial cameras are arranged in a circle above the pallet, which are used in conjunction with the visual light source to collect images of the bevel area for processing

Configuration parameters

Advantages of ai algorithm

• Accuracy of results: Compared with traditional vision algorithms, the stability of deep learning algorithms is greatly improved, and it can adapt to general disturbances in the environment and background. At the same time, the accuracy of the algorithm is also higher than that of traditional vision algorithms;
• Algorithm versatility: For different defects, only a small number of defect samples need to be collected. After sufficient training, different defect samples can be automatically identified. The algorithm adopted is a unified algorithm framework;
• Development timeliness: Since there is no need to develop algorithms for different defects, the entire development cycle is greatly shortened, and general visual defect projects only need 2-3 hours to go online for testing;
• Environmental reliability: Due to the guarantee of GPU computing power, the entire system can work in a high-temperature environment for a long time with sufficient computing power margin.

Advantages of big data cloud platform

• The model has wide applicability: by distinguishing whether there is a template or not, it can basically cover the detection requirements of surface defects, and new defect models can be added arbitrarily;
• The robustness of the model is good: through the system signal consistency adjustment, it is basically guaranteed that there is no need for fine-tuning on site, and it can run normally when it goes online;
• Fewer training samples are required: In the case where a large number of defect samples cannot be obtained, the automatic synthesis of samples can be used to meet the needs of deep learning with large data volume training samples;
• Fast model reasoning speed: Control the cost of model deployment, and provide customers with a cost-effective deployment platform as much as possible on the premise of meeting the needs.