Four cameras become one view of the ship from above.
AegisView AI stitches four fisheye cameras into one overhead view of everything around a vessel.
In procurementas of In procurement for a Korea Coast Guard program, through partner J3C
AegisView AI
- Languages
- None · images and video
- Also known as
- AVM
Four hull cameras merge into a single view from above.
- CAM 1 · bow
- CAM 2 · starboard
- CAM 3 · stern
- CAM 4 · port
- Camera field of view
- Stitched overhead view
AegisView AI shows the water around the ship. The crew decides and steers.
What it handles
- Takes in four fisheye camera feeds.
- Stitches them into one overhead view around the vessel.
- Runs on a PyTorch pipeline on GPUs.
- Comes with a camera-placement design tool for each vessel.
What it hands to a person
- Steering, berthing and every navigation call stay with the crew.
- Lookout stays with the watch. The view supports it.
- Reading the view and acting on it is the crew's job.
AegisView AI is a PyTorch pipeline that runs on GPUs.
Specification
- Languages
- None: it works on images and video, not speech
- Integrations
- Four fisheye cameras · camera-placement design tool
- Deployment
- On-premises GPU
- Data handling
- Installed on an on-premises GPU, footage is processed on your own servers. Any footage on this site is labeled with vessel or site, date and camera
- Pipeline
- PyTorch on GPU
AegisView AI is quoted per project. Here is what sets the price.
Open the pricing sheetWhat sets the price
- Cameras
- Vessels
- On-premises GPU
- License term
Five questions buyers ask about AegisView AI, answered plainly.
Is AegisView AI the same as AVM?
Yes. AegisView AI has also been presented as AVM, short for around-view monitoring. It stitches four fisheye cameras mounted around the hull into a single overhead view of the water around the vessel. This page is its home on the web.
How many cameras does it use?
Four fisheye cameras, one each at the bow, starboard, stern and port, each covering one side of the vessel. Their views are stitched into one overhead frame. Exact mounting positions for each vessel are worked out with the camera-placement design tool and agreed with the operator.
Can I see real footage?
Not yet. The system is in procurement, so footage is published once the terms allow it. When it is, every clip will be labeled with vessel, date and camera, and shown raw next to the stitched view. Until then, this page shows the layout drawing only.
Where is the video processed?
By a PyTorch pipeline running on GPUs. Installed on on-premises GPUs, footage is processed on servers you control. The GPU setup is sized to the project in the working session, and processing latency is published with its date once measured.
What does AegisView AI cost?
It is quoted per project. Four things set the price: the number of cameras, the number of vessels, the on-premises GPU setup and the license term. Public-sector projects depend on their own terms, so we list the cost drivers here instead of a price.