| I&T Solution |
AI-Enabled Automated Vessel Air-Draught Detection
(REF: S-1564) |
| Trial Project |
|
| Solution Feature |
- Image Segmentation and Object Detection Models trained with large dataset under various weather conditions to recognize highest point and waterline for maximum accuracy.
- Incorporate image preprocessing techniques (adaptive thresholding, noise reduction, and histogram equalization) to enhance visibility of vessel's highest point and waterline.
- AI module designed to handle static images and video streams, processing them in real-time so as to provide measurements quickly, adhering to the 2-second requirement.
- Training dataset will be augmented by variations such as rotations, zooms, and brightness adjustments to generate additional training images with synthetic data generation.
- The AI module will be wrapped with API endpoints, making it easier to integrate with other systems or applications if needed.
|
| Trial Application and Expected Outcome |
- Feed images/videos captured under various weather conditions (fog, rain, and bright sunlight) to operate accurately regardless of weather conditions.
- Low-Light/Night-Time Operation: Confirmation of the system's capability to function effectively in low visibility. This ensures the port can safely operate around the clock.
- Disturbance Handling Test: Feed the system images/videos with disturbances like water splashes, birds, or other temporary obstructions.
- Compare the results to showcase the reliability of the automated process against the traditional manual approach.
- Validate the system's speed and efficiency in a real-world setting in real time to ensure that the system can handle live operations seamlessly.
|
| Additional Solution Information |
AI-Enabled Automated Vessel Air-Draught Detection Proposal by Beyond Limits - 20230906.pdf
|
| Info on I&T Solution Provider |
| Solution Provider | : | BEYOND LIMITS HONG KONG LIMITED | | Address | : | Unit 336, 19W, | | Contact Person | : | Hilda Li |
| Position | : | Technical Sales Director | | Tel | : | 93160397 | | Email | : |
hli@beyond.ai | | Webpage | : | beyond.ai |
|
