Container Body Damage Detection

With the rapid expansion of the container industry, shipping carriers and port operators are facing growing demands for container damage inspection. Severe container defects may result in cargo damage and loss, and even pose safety hazards to container trucks during transit.Seizing opportunities brought by intelligent port upgrading, Jingwei HiRain has developed a container damage detection system based on laser sensor data and machine learning algorithms, helping ports improve transportation safety and operational efficiency.

Composition of the Container Body Damage Detection System

The container damage detection system consists of four core modules: perception module, algorithm module, edge computing unit and data interaction module.

Optical cameras and LiDAR in the perception module collect container image data and point cloud data respectively, then transmit such data to the edge computing unit via Ethernet cables. After receiving the data, the edge computing unit processes and computes data with various algorithm models embedded in the algorithm module to generate final container damage detection results.

Preloaded with multiple algorithm models, the algorithm module covers image recognition & processing model, point cloud trigger acquisition model and point cloud feature extraction model. Container damage identification is ultimately realized relying on machine learning algorithms.

Complying with specified communication protocols, the data interaction module establishes one-way or two-way communication links. It uploads identification results to the enterprise cloud platform and delivers processed sensing data for subsequent model training and iteration.

Schematic Diagram of Container Inspection Point Cloud (Front/Rear View)

Features of the Container Body Damage Detection System

1.The sensor boasts a large vertical Field of View (FOV), delivering excellent detection performance for container side panels.

2.Automatically identifies and extracts point cloud data within designated regions.

3.Adopts data-driven machine learning algorithms featuring fast processing speed and support for iterative algorithm upgrading.

4.High product integration enables convenient on-site installation and deployment.