Wildbook’s Wild Year in Review 2018
Happy New Year!
From all of us at Wild Me and Wildbook, have a wonderful new solar year! May it be filled with discoveries, wonders, and the beauty of our planet. May it be full of zebras and turtles and sharks and lynxes and whales and elephants and all the other creatures big and small!
For the new year, give a gift of conservation.
What a year! 2018 was an exciting period of incredible growth and achievement across the Wildbook project (wildbook.org) and its non-profit engineering home Wild Me (wildme.org). With a strong focus on Artificial Intelligence, 2018 catapulted us to the bleeding edge of conservation technology and citizen science. From mysterious Bitcoin donors to big partnerships with Microsoft, Vulcan, BOEM, NOAA, KWS, and others, 2018 has shown that there is a lot of need and support for AI for conservation. We need to combat extinction faster, and we need to engage the public in our efforts. Wild Me’s advanced use of AI is hitting its stride and reaping the benefits. We supported hundreds of scientists, engaged thousands of citizen scientists, and tracked tens of thousands of animals across the globe in 2018! Read on to find out the details.
New and Noteworthy in 2018
- Non-human collaborators?!?! Wildbook added an “Intelligent Agent” to our Wildbook for Whale Sharks project (whaleshark.org), creating a blended human and A.I. environment in which an A.I. collaborator is finding new data for human researchers and curating data for them. Checkout “Unveiling ‘Intelligent Agents’” below.
- IoT…means the “Internet of Turtles”, right? https://iot.wildbook.org is now in beta, allowing researchers for Hawksbill and green sea turtles to test computer vision and identify individuals based on the scutes on their heads. More species support and a full release coming in early 2019!
- Dolphins can be ID’s with their fins! Working closely with PhD candidate Hendrik Weideman at RPI and the Sarasota Dolphin Research Project, we have new computer vision for bottlenose dolphins deployed and under testing on Flukebook.org.
A mysterious Bitcoin donation from the Pineapple Fund in January allowed us to hire long-time collaborator Jason Parham full time, adding research-level computer vision and machine learning talent in-house to Wild Me’s existing team of three conservation software engineers.
Wildbook enabled a historic census of the endangered Grevy’s zebra and, for the first time ever, reticulated giraffe in Kenya. is the most accurate species census and it is done by ordinary people — school kids, park rangers, tourists… anybody! — taking photographs of animals for two days in January. More than a 1000 people took more than 50,000 pictures! Wildbook made it possible to identify every zebra and giraffe and provide the most accurate count ever: about 2800 Grevy’s zebra and 2300 reticulated giraffe. The Rally will become a biannual event, with the next edition in January 2020.
In February, Dr. Alex Dehgan joined Wild Me’s Board of Directors. Alex brings the wisdom of a long career of academic, government, and non-profit work in conservation, innovation, and Ecology. We’re excited to have him helping us through a big year of change!
In March, NVIDIA featured Wildbook’s use of NVIDIA GPUs for deep learning at the GTC conference in San Jose and in its “I am A.I.” series of videos. Check out their great video of how Wildbook works. With over 10,000 views, the message is getting out about how Wildbook applies A.I. to help wildlife researchers understand populations and combat extinction!
In June, Wild Me joined Microsoft’s A.I. for Earth program as a flagship project, allowing us to grow and consolidate into a focused, full-time team of engineers bringing A.I. to wildlife conservation. The A.I. for Earth program also helped us bring long-time citizen scientist, fundraiser, and Director Jason Holmberg on as Wild Me’s first Executive Director in June 2018, providing focused leadership for our engineering team and outreach efforts. More coverage of our partnership with Microsoft will be unveiled in 2019!
In August, Wild Me signed a contract to bring computer vision to North Atlantic right whale research in partnership with NOAA. We’re excited to be moving forward with initial computer vision work in early 2019.
In November, National Geographic covered Wild Me’s leadership in applying A.I. to wildlife conservation<, presenting how Wild Me combines multiple forms of machine learning (computer vision, natural language processing, machine translation, etc.) to find useful population data in social media posts that are otherwise missed by the research community.
Also in November, Wild Me partnered with Vulcan, Inc. on a new project to apply machine learning to modernize aerial elephant census work. Combined with the ongoing research in individual elephant photo ID conducted by our partners at Rensselaer Polytechnic Institute and with the support of h2o.ai, we see a long future of elephant conservation work ahead for Wild Me!
Unveiling “Intelligent Agents”
Seeking to add beneficial artificial intelligence to all Wildbook-based communities, Wild Me released an “intelligent agent” in 2018 that applies multiple forms of machine learning (e.g., computer vision, natural language processing, machine translation) to transform social media-sourced data into curated data for whale shark researchers. The agent datamines YouTube nightly for real world whale shark sightings in the wild and interacts with the video poster when data is missing. For the whaleshark.org community, this has resulted in a free flood of new data (a 100+% increase) and our first non-human participant in the 180+ researchers collaborating through whaleshark.org, the Wildbook for Whale Sharks.
Our Intelligent Agent has exceeded our expectations:
- 91% of data was not duplicated by the human research community, indicating that that agent is a unique data collector using a unique data source (public social media)
- The agent engaged over 2200 video posters on YouTube in the last year, finding an audience largely missed by our traditional outreach efforts. Compare this to the 2000+ citizen scientists that submitted data across all Wildbooks and the GGR2 in 2018. Intelligent Agents engage more people annually than we can engage through traditional outreach.
- The agent outpaces human researchers each month in the volume of whale shark sighting data collected and the number of sharks individually identified. It is a .highly effective non-human collaborator within a global, online research community of 180+ human researchers and over 6000 contributors from the public.
Want to learn more? Check out David Carmon, General Manager of AI Marketing at the Microsoft Ignite conference presenting how an Intelligent Agent works: https://youtu.be/9vvzI9gQnjE?t=3155
By the Numbers: Metrics for Success
The numbers don’t lie. 2018 was a record success. Here are a few of our statistics:
Number of Wildbook full-time engineers: 4
Number of researchers served across Wildbooks: 400+
Number of citizen scientists engaged through direct submission to a Wildbook: 2000+
Number of data contributors from social media engaged by our Intelligent Agent FOR WHALE SHARKS ONLY: 2200+
Number of new animals identified individually with photo-ID and computer vision: 3600+
Number of new wildlife sightings in Wildbook: 23000+
Image detections and comparisons made per day across Wildbooks using ML and computer vision: 1.5 million
Can you help us make 2019 an even greater year, adding more A.I. to the conservation fight and engaging more members of the public in our work?
Please support AI and humans fighting extinction together with Wild Me and Wildbook!