Automatic detection of contaminants on sea container exteriors to improve detection of unwanted exotic organisms

The global movement of sea freight provides a significant pathway for the movement of and introduction of unwanted exotic organisms. Along with the potential economic and environmental impacts, there is loss of biodiversity for the receiving country. While risk analysis and profiling; offshore cleaning stations; and visual inspection at the border during unloading mitigate the risk of introduction, it is time and labour intensive, with a high risk of slippage. The ability to rapidly survey a larger proportion of sea containers for external contaminants to both improve detection of unwanted organisms and reduce biosecurity risk is being addressed with research to develop semiautomated container scanning tools for seaports. This is based on two key technologies: a scanning system using cameras in conjunction with edge-applied neural networks, and laser-based topology scanning. Two pilot studies using cameras demonstrated that a deep-learning neural network could be used to successfully identify externally contaminated sea containers. Current research aims to extend the pilot studies to improve the robustness of detection systems, whilst also developing an imaging and computing system that would allow for the system to be deployed in a port environment. The laser-based topology scanning technology will lead to an anomaly detection system, based on the identification of deviation from the expected smooth and largely uniform surface topography of a shipping container. Deviation may indicate the presence of a contaminant. By selecting suitable wavelengths, the system can also use spectral information to maximum the reflectance difference between contamination and container surface. The advantage of topography scanning laser approach is that it is much less computation intensive as a simple threshold can be used to identify the possible contamination based on roughness, reflectivity, and protrusion in real time. Results to date show that there are consistent spectral differences between container and contaminations, particularly around 1450 nm where contaminations such as insects and soil have a lower reflectance due to absorbance caused by the presence of water. Results from studies using the two technologies will be presented, along with an examination of potential challenges to the application and suitability of these systems in a seaport environment. The need to improve detection of unwanted organisms in a sea port environment makes the use of these technologies vital if biosecurity risk is to be reduced.