AI Deep Learning Based Reagent Bottle Online Visual Inspection System

AI Deep Learning Based Reagent Bottle Online Visual Inspection System

Company Profile

Anhui Keye Intelligent Technology Co., Ltd. is a Chinese national high-tech enterprise specializing in the application of AI technology in visual inspection industry. As a leading manufacturer&Designer, It can provide the most advanced level AI visual inspection equipment and the customized AI system solutions of visual imaging, computing power, algorithm, and automatic control software for various sub-industries.

The company was established in 2011, it has a high-level R&D team from top domestic optical research institutes and a core doctoral team from the USTC, covering the entire technology chain of optics, electrics, mechanics, algorithm and software. Our products can be applied in beverage, pharmacy, dairy, wine, textile, new energy, agriculture and other packaging industries.

Detection principle

The system transports the bottles to the predetermined station in a stable and orderly manner, and then transmits the captured sample image to the image processing software through the high-pixel industrial camera, compares with the standardized sample and distinguishes the defective product and the good product, and controls the corresponding pneumatic components to reject defective products and collect qualified products.

The software sets the comparison standard for the samples, operators can control the comparison precision and choose the precision standard that best suits themselves, thereby controlling the defect rate and greatly reduce costs.

Detection details
Our design plan uses 6~8 sets of high-pixel industrial cameras and high-performance stroboscopic light sources to carry out comprehensive detection for bottle mouth, bottle bottom and bottle body defects. The whole set of detection equipment includes mechanical parts, visual electronic hardware and detection systems. The software, the entrance and exit of the detection machine can be integrated with the conveyor line, the detection speed is 100-200pcs/min, the equipment can realize 7*24 hours of operation, and the unqualified products can be automatically rejected.

Operation Parameters

Configuration List

Hardware:

l Camera: We use our own developed camera (the chip is from Japan), which has clearer image recognition and is more compatible with our own software.

l Light source: We have a dedicated optical team to develop various light sources to adapt to the optical scene requirements of different industrial scenes.

l Embedded computer: The KEYE team independently developed the embedded computer NPU computing box, which replaced the traditional PLC industrial computer; it is not afraid of sudden power outages and high temperature scenes, and can better adapt to the high temperature environment.

Software:

1. Algorithm: We use the latest AI deep learning algorithm. Most products only need a small amount of sample learning to complete a new product model; the machine can continuously learn some defects and good samples like the human brain, making the detection effect better and better.

2. Software: The latest KEYE V16.0 software can realize a variety of data statistics and human-computer interaction functions, and the touch screen operation is simpler.

3. KEYE cloud platform: The AI cloud platform can create its own cloud platform account for each customer, and the pictures of each training operation will be saved in its own account; even if the algorithm is lost due to software algorithms or other problems, all previous data can be retrieved through an independent cloud platform account. (KEYE team has technicians who regularly maintain this platform).

4. After-sales service: Provide long-term remote service, customers only need to keep the device in good network connection.

About Us

Development History

The technology center is composed of algorithm department, software department, optics department, hardware department and mechanical department. The algorithm team currently has 7 PhDs, ensuring that the AI algorithm has the ability to continuously update.

Organization Structure