To address industry pain points such as low efficiency in manual container inspections at ports and yards, frequent missed or false detections, and imprecise damage localization, Fuwei Intelligent has developed a proprietary container damage detection system. Leveraging multi-angle high-definition imaging and deep learning AI algorithms, the system achieves fully automated, non-contact detection of damage across the entire container surface. It provides high-precision quantitative assessment using a 3cm × 3cm grid and is designed to support high-speed throughput operations at terminals.
I. Multi-camera panoramic imaging for comprehensive coverage
The imaging hardware is arranged in zones to fully cover five sides of the container:
1. Left, right, and top sides: Three line-scan cameras capture seamless panoramic images without distortion or stitching artifacts.
2. Front and rear ends (door areas): Two global-shutter area-scan cameras mitigate interference from water stains, reflections, and motion blur.
The equipment features high IP67/IP66 protection ratings, ensuring reliable all-weather outdoor operation despite port conditions such as salt spray, light rain, and strong backlighting.
II. Deep learning AI algorithms for precise identification of seven damage categories
The system employs a dual-layer neural network combining detection and pixel segmentation, integrated with the CBAM attention mechanism. It is specifically optimized to detect seven types of container damage: rust, dents, scratches, holes, and weld cracks.
- Standard detection unit: 3cm × 3cm; damage localization error: ≤ ±5cm;
- Overall recognition rate: ≥ 90%; missed and false detection rates: < 5%;
- Stable image capture at vehicle speeds of 0–15 km/h (constant speed or with minor acceleration/deceleration).

III. Core operational performance meeting high terminal throughput demands
1. Efficiency: Total time from capture to result generation is ≤ 10 seconds per container; hourly throughput capacity is ≥ 300 containers;
2. Stability: Supports 24/7 continuous operation with an MTBF (Mean Time Between Failures) of ≥ 10,000 hours;
3. Environmental compatibility: Operates in lighting conditions ranging from 100 lux to 100,000 lux, as well as in light rain and foggy weather;
4. Output: Automatically maps damage using a grid overlay and color-codes it into three severity levels: minor, moderate, and severe.
IV. Lightweight software functionality balancing automated inspection with manual operations
1. Panoramic grid visualization: Overlays a standard grid onto the panoramic container view, allowing for localized zoom-ins to verify damage details;
2. Manual editing and traceability: Supports manual addition, deletion, and modification of damage annotations, with full operation logs retained for traceability;
3. Comprehensive data management: Locally stores container images and inspection reports for 180 days, with automated cleanup alerts upon expiration;
4. Convenient integration and export: Standard HTTP interfaces connect to container management and TMS systems; supports one-click export of JPEG images and Excel statistical reports;
5. Historical comparison and retrieval: Enables rapid searches by container ID, time, or damage type to compare changes in container condition over time.
V. Standardized acceptance for worry-free deployment
The Fuwei container damage inspection system comes with comprehensive acceptance criteria: 72 hours of continuous, full-load operation without system failure; stable performance across various weather conditions and operating speeds; and mandatory compliance with all metrics regarding recognition accuracy, efficiency, and software functionality prior to delivery. Standardized reference images of damage types are provided to assist with on-site maintenance and manual verification.
Compatible with 20ft, 40ft, and 45ft dry containers, reefers, and flat-rack containers (suitable for both steel and aluminum bodies), this system offers a one-stop solution to the challenges faced by port yards—namely the high costs, low efficiency, and inconsistent standards associated with manual container damage inspection.