Marine AI has released a new suite of sophisticated software plugins for the company’s Guardian Vision platform, which provides computer vision-based sensor analytics for ocean-going vessels such as USVs (unmanned surface vessels). The plugins have undergone significant on-water verification and testing, and are now available for users to enhance safety and situational awareness for their marine and maritime applications.
Built around powerful IBM and NVIDIA deep learning and edge computing technologies, Marine AI’s Guardian software uses artificial intelligence-based object identification and hazard analysis to aid manned and unmanned vessel control and safety management. Deploying this state-of-the-art capability at the edge allows all decision-making computation to be performed on board, vastly reducing the amount of satellite bandwidth typically required for this work to be performed remotely.
Guardian Vision provides multi-platform, scalable, TLS-encrypted secure computer vision capabilities that can be deployed on a variety of vessels to provide real-time risk assessment and actionable information transmission to land-based teams. The new plugins, which enhance the effectiveness of the system even further, include:
- Waterline Detection and Correction
- Low Light Enhancement
- Digital Video Stabilisation
- Object Direction Estimation
- Object Detection and Labelling
- Object Classification
- Object Tracking
- LOMO (Low observable marine objects) Detection
The two images below give an example of the effectiveness of applying the Low Light Enhancement plugin:
Marine AI’s state-of-the-art Guardian software has been utilized by both the MAS (Mayflower Autonomous Ship) and the UK Royal Navy’s XLUUV (extra-large unmanned underwater vehicle). The company invites interested parties who think that Guardian computer vision capabilities can be an asset to their marine and maritime projects to get in touch.