Just asking for input because I’ve never actually done a spotting and reconnaissance project.
Industry Backstory
Whale watching is big business in California, generating millions of dollars yearly from less millions of customers. Seven days a week, cephalopod-seeking sightseeing ships regularly depart with hundreds of tourists, stuffed to the gills like sardines. Typically they charge $20 or more for the 2-3 hour excursion, however since there can be no guarantee of an encounter on any given day, on those days when nothing is found, rain checks must be issued, leaving the operator on the hook for potentially thousands of dollars per outing. Current whale spotting methods are labor intensive and somewhat hit or miss. Therefore, there is quite a bounty to be paid to the surveillence installer who can rig up a decent automated whale detection system for under $100,000.
Goal:
To install an autonomous, self-learning, whale detection and notification system that can supply gps coordinates, pod direction and pod size information in real-time to tour boats, docked or underway, directing them to the closest active sighting.
Solution:
I’m thinking that a typical three camera and VA capable NVR, setup off the dock, should be sufficient to handle this:
1. Wide angle, high-resolution multi-imager, 180 or 270 degree.
2. 4K@30 30X zoom PTZ
3. Thermal zoom PTZ
The basic workflow would be camera #1, the overview camera, would be setup to trigger on motion, though because of its wide view wouldn’t be able to identify the object. It would therefore trigger camera #2, the 4K PTZ to zoom to the location and collect frames. The thermal ptz, camera #3, would also pan/zoom to the location, looking for a mammalian heat signature.
The information would then be passed off to the NVR’s object detection algorithm, which has been pre-trained for whales identification using “deep-sea learning”. If a match occurs, gps coordinates would be derived from absolute ptz coordinates using some really fancy math. Then the sighting information and coordinates, along with a snapshot would be sent via mobile delivery to the tour captains.
Optional:
To eliminate false alarms, a drone could be dispatched to location by the NVR for final verification.
Comment:
Does this plan sound reasonable? 2 man job?
Any input welcome, please feel free to spout off. Thanks!
Posted: 04/01 11:15 PM local time