In the 2020 AI for Earth Summit, an annual summit hosted by Microsoft to showcase the research by AI for Earth grantees, a new repository of computer vision datasets for biology and conservation was presented.
The LILA BC is an online and open-access repository that holds hundreds of computer vision datasets for researchers that want to harness the power and capabilities of computer vision for biology and conservation.
As I wrote in my Fish Classification Datasets blog, one of the main challenges of finding datasets for computer vision tasks, is that the objects within images need to be labelled and fully identified. The LILA database solves this issue by hosting fully labelled datasets in a variety of modalities, and they currently host over 10 million labelled images.
The datasets can be divided into four main categories:
Camera trap datasets: sequences of camera trap images from various monitoring networks around the world.
Aerial surveys: images from unnamed aerial vehicles and sensors.
Geospatial datasets: satellites images with land-cover labels.
Bioacoustic datasets: recording of various marine mammal species.
Here I highlight three of the most diverse and largest datasets
1.4 million camera traps images from 12 countries. The most diverse camera trap dataset in LILA
3.7 million camera trap images from 5 locations in the United States
Manuscript: Tabak, M. A., Norouzzadeh, M. S., Wolfson, D. W., Sweeney, S. J., VerCauteren, K. C., Snow, N. P., … & Teton, B. (2018). Machine learning to classify animal species in camera trap images: applications in ecology. Methods in Ecology and Evolution.
More than 500k Sentinel images with patch-level land cover labels acquired between June 2017 and May 2018 over 10 European countries.
For more information and to view all the datasets please visit http://lila.science/
Follow the progress of FishID; an automatic platform for fish species identification and abundance quantification through this blog or @seabassphd.
Sebastian Lopez (Seabass), is a PhD Candidate at the Australian Rivers Institute where he is developing and applying artificial intelligence tools to monitor fish populations in marine ecosystems.