| I&T Solution |
Equipment Gauge Reading Recognition by AI
(REF: S-1668) |
| Trial Project |
|
| Solution Feature |
- Automatically recognize and record the readings of equipment gauges, including analog gauges, rotary meters with needles, digital meters with LCD displays, and seven segment displays, by using target detection, OCR, CV and other AI technologies.
- Generate a unique QR code for each equipment gauge, which contains information such as device name, type, serial number and location etc.
- Establish a knowledge base for each type of instrument, and store the normal reading range of the device, which is used for subsequent analysis of various readings to determine whether there are abnormal cases.
- Scanning the QR code and readings by the MOBILE APP. The AI engine will analyze and save the recognized results in database. The stored information includes readings, the QR code, scanned picture, time etc. Multiple readings for a single equipment can be recognized at the same time when necessary.
- The data will be transmitted to the existing system platform in encrypted form according to the interface protocol and data format of the existing system platform. A web portal is provided for the generation of reports.
|
| Trial Application and Expected Outcome |
- The meter detection module comprehensively and accurately analyzes the image through deep learning algorithms, and is able to quickly and accurately locate the position of the meter in the picture, providing a reliable basis for subsequent processing.
- The scale positioning module further refines the recognition process and accurately locks the scale line on the meter, providing accurate data support for the subsequent pointer recognition and reading calculation.
- The pointer recognition module adopts advanced algorithms such as convolutional neural network (CNN) in deep learning, which can efficiently recognize the position and angle of the pointer on the meter, so as to accurately obtain the current state of the meter.
- Intelligent error correction module: By configuring the knowledge base of various instrument equipment, the real-time collected instrument readings are analyzed by comparing with the knowledge base, so normal readings and abnormal readings are distinguished.
- The recognition process of the measuring instrument recognition algorithm: obtaining images, detecting the instrument, performing type judgment, scale positioning, pointer positioning, relationship calculation, comprehensive calculation, and calculating results.
|
| Additional Solution Information |
設備讀數AI識別方案 (技術方案）.pdf
|
| Info on I&T Solution Provider |
|
