|Ol Pejeta Conservancy||Next Challenge: How can our customer service culture meet customers' changing desires?|
**NOTE** The grand prize winner will be offered a flight (economy and capped at $1,000) to Kenya, transfers to Ol Pejeta, and 7 nights food and accommodation in a luxury lodge with a VIP tour of the conservancy and the chance to work on putting their algorithm into action on the ground. (Over $4,000 total value).
Ol Pejeta has 110 endangered rhino spread across 90,000 acres as well as a number of other endangered species. Protecting and growing the numbers of these animals over such a wide area represents a major logistical challenge. Whether it be preventing poaching, or simply monitoring populations, current solutions are highly manpower intensive and risky.
There are a number of tracking technologies out there, including UHF tags in rhino horn and other such transmitter based techniques but most share the same challenges:
1. You have to drug the animal for attachment which is dangerous for both humans and the health of the rhino
2. You have to find an effective way to actually attach the sensor so that it remains attached indefinitely, doesn’t impinge on the rhino’s natural behavior (which includes rubbing up against trees) and isn’t excessively visible to tourists.
3. You need to have a chip/sensor system with an acceptable range so as to easily cover a wide area, and so as to minimise human proximity to the wildlife
4. All solutions to date have therefore required an active sensor, which requires battery power and lasts for a maximum of 1-2 years. This is a problem because you don’t want to have to continually re-drug animals to change batteries.
Ol Pejeta believe that the solution has to be found in taking a wider, indeed birds-eye view. Step one has been a successful fundraising program on Indiegogo, which has raised funds for a new aerial drone. The drone has both visual and thermal imaging cameras that give us the ability to cover large swathes of land in short spaces of time.
Step two we believe will come with significant advances in recognition software. We believe that the future has to be software that can use a combination of visual and thermal signatures to identify wildlife automatically. This would bypass the attachment and tracking hurdles, and revolutionize the conservation world.
The most immediate application for such a capability would be to transform the process of taking wildlife censuses. At present these are done infrequently (in Ol Pejeta's case once per year), by light aircraft, with one km wide transects, and using the mark one human eyeball! It is costly and can be inaccurate. Instead, with a drone flying tight transects and recording HD footage for later review and analysis, censuses could be taken monthly or even more often and with far higher levels of accuracy.
Basic: Create an algorithm that automatically recognize the signature of a life-form (animal or human) and ‘bracket it’ on a user interface for the operator to view/review.
Intermediate: Take the basic level a step further and distinguish between human and animal life forms.
Advanced: Take the intermediate level a step further and be able to identify specific species. To start with we would want to focus on elephant and rhinos.
Solvers can focus on any of the three levels of difficulty (basic, intermediate, advanced), but solutions to the advanced deliverable are most preferred. Please demonstrate your algorithm through some kind of user interface (basic is fine) where any visual/thermal imagery running through the user interface automatically flags up life-forms.
Ultimately the algorithm is king. If we need to re-spec our drones with different hardware and data transmission systems that will better enable the algorithm, we would certainly do that. However, for now we have example footage. (see attached materials & links below)
Some useful pointers on physiology of rhinoceros that remains fixed:
1) Ear notches 2) Gap between base of horns 3) Sex 4) Size (change with age but only slowly) 5) Horn size/shape (change with age but only slowly)
The grand prize winner will be offered a flight (economy and capped at $1,000) to Kenya, transfers to Ol Pejeta and 7 nights in a luxury lodge with a VIP tour of the conservancy and the chance to work on putting their algorithm into action on the ground.
The two runners-up will receive a cash-reward of $500.