By Real-Time Monitoring and Comprehensive Identification Rate up to 98%, AI Technology Makes Guangzhou Port More Intelligent in Tallying

Recently, the AI+OCR (Optical Symbol Recognition) artificial intelligence tallying research and development project of GOST Nansha Phase III Container Terminal has been successfully tested as qualified in the internal acceptance meeting. According to the GPG's request to upgrade the OCR image capture recognition technology of the original quay crane to artificial intelligence AI+OCR video stream recognition technology, the project head from GOST and working team upgraded the container OCR image recognition technology algorithm to artificial intelligence AI deep learning algorithm, thereby greatly improving the recognition rate and stability of container number, vehicle number, container position, etc., with a comprehensive recognition rate of 98%.

Influenced by the COVID-19 epidemic, all the staff of GOST Hongbiao Innovation Studio and Nansha Phase III tallying department overcame difficulties and made every effort to push forward the research and development progress of AI artificial intelligence tallying technology upgrade project, by giving full play to the exemplary role of CPC members, and proposing a total of more than 100 modification suggestions to the cooperating partners during the operation. The team overcame the impact of the COVID-19 epidemic. After 81 days, serving 223 shifts and 200 voyages, the team tested 35,912 containers under various climatic conditions with the system, and the results prove the system function has reached the designed standard. The new system is characterized by "real-time monitoring, automatic identification, automatic matching, automatic release, error detection and interception, automatic transmission" and meets the overall requirements of "advancement, stability and replicability" by the GPG.

During the upgrading technology research and development and equipment renovation, the project team has independently developed the automatic calculation technology of loading and stowage position and obtained the national utility model patent, on the basis of inheriting the original automatic identification technology. The project team innovatively adopted the automatic pushing technology of suspected damaged containers, thereby realizing the two innovative functions: "ground-to-air integration" AI+OCR identification, and automatic identification of damaged containers, and breaking through the bottleneck in the industry as the first solution for such problems nation-wide. Such intelligent tallying technology has overturned the traditional tallying model where a tallyman operates abridge crane, allowed a tallyman to operate many lines in the background, effectively improving the working environment, reducing labor intensity and labor costs; in addition, it is capable of effectively improving the terminal operating efficiency, thus providing technical support for automated and remotely controlled bridge crane in Nansha Phase III and fully automated bridge crane in Nansha Phase IV in the future. (Wang Hailang and Wen Zhihuan)