Jiaxun Feihong's wholly-owned subsidiary, CASTELiTS, has launched a fully automatic container empty box detection system, offering a new experience in intelligent detection empowered by AI.

2025-04-07

Source：Jiaxun Feihong official Weibo

At this critical stage of AI - driven intelligent transformation of customs, the continuous expansion of global trade has posed unprecedented challenges to the efficiency of port supervision. The traditional container inspection model, which relies on manual judgment, has shown obvious limitations under the dual pressures of surging throughput and increasing requirements for customs clearance timeliness.

Beijing Jiaxun Feihong Electrical Co., Ltd. (Stock Code: 300213) and its wholly - owned subsidiary Shenzhen Hangtong Intelligent Technology Co., Ltd. (hereinafter referred to as "CASTELiTS") have innovatively launched a fully - automatic container empty - box detection system based on multimodal AI large - model and intelligent agent (Agent) collaborative decision - making, leveraging years of technical accumulation. This breakthrough technology redefines the standards of intelligent detection and injects new momentum into the digital transformation of customs supervision.

The system has the following core advantages:

  • Multimodal fusion perception with detection accuracy of 99.7%

The system integrates advanced sensing technologies such as LiDAR point cloud modeling and high-precision visible-light 3D reconstruction to create a "digital twin" of the container. Driven by large-model-based feature matching algorithms, it can accurately identify microscopic residues inside standard containers ranging from 20 to 45 feet, achieving a groundbreaking detection accuracy of 99.7%. Even in complex scenarios with extreme lighting, rain and fog, or container deformation, it can still maintain centimeter-level recognition accuracy, significantly improving the efficiency of customs supervision.

1909120551287156738.png

Intelligent Agent Autonomous Decision - Making, Millisecond - Level Risk Assessment

The system's intelligent decision - making core, based on a multi - agent architecture, achieves full - process automated detection through the collaboration of multiple agents for detection, re - checking, and evidence - keeping.

  • Detection Agent: It autonomously coordinates the robotic arm to dynamically adjust the scanning strategy to fit various operation scenarios such as passage through checkpoints.

  • Re - check Agent: When a suspicious container is detected, it automatically triggers a multi - spectral re - inspection process to ensure the reliability of the results.

  • Evidence - keeping Agent: It generates tamper - proof electronic reports to achieve full - process traceability.

Continuous Evolution, Building Domain - Specific Knowledge Graphs

The system uses a reinforcement learning framework and continuously optimizes based on large - scale empty - container detection data:

  • It has been adapted to 13 standard container specifications worldwide;

  • It can accurately identify hidden layers, residues, and other abnormal conditions;

  • It supports cross - port knowledge sharing through federated learning, reducing the false - detection rate to three in a thousand.

The system has been successfully deployed in several key ports in China and continues to optimize the efficiency of smart port services, significantly improving the efficiency of cargo customs clearance. In the future, CASTELiTS will continue to adhere to the development concept of innovation - driven, committed to the innovative application of artificial intelligence technology in the field of port management, deepen the implementation of artificial intelligence technology in more scenarios, and continuously explore new paths for the construction of smart ports.

Related Materials