🌳🚁 Revolutionary drone workflow captures ultra-HD photos of tropical tree canopies! Our new method captures 300 trees/day from 6m above, helping botanists ID species that are impossible to see from the forest floor.
📄 doi.org/10.1101/2025.09.02.6…
We’re releasing SelvaBox, the largest tropical tree detection dataset from drone imagery. Our models trained on SelvaBox achieve competitive zero-shot detection performance on unseen tropical tree crown datasets, matching or exceeding competing methods. arxiv.org/abs/2507.00170
We're working on Barro Colorado Island with @HMullerLandau's team to refine our winning #XPRIZERainforest canopy tree mapping solution. We're now confident enough in our workflow to the drone move closer to trees to take even better photos. #BioCLIP correctly ID'ed that one!
ALT Close-up photo taken at 6 m above the crown using the telephoto camera of the DJI Mavic 3 E drone. BioCLIP correctly identified this image as Ficus insipidia: https://panamabiota.org/stri/taxa/index.php?taxauthid=1&taxon=64890&clid=180
🥇We’re thrilled to announce that our @LimelightRF team has won the #XPRIZERainforest competition! 🏆
🌳We have developed the canopy tree mapping solution using drone imagery and AI 🌴
Follow 🧵below for details!
So happy to announce that our team @LimelightRF won the #XPRIZERainforest competition! We delivered a canopy tree mapping solution using drones and AI that can transform tropical forest biodiversity monitoring. Thanks @IVADO_Qc for the support.
We’re so thrilled to announce the winners of the #XPRIZERainforest competition!
🥇Limelight Rainforest
🥈Map of Life Rapid Assessments
🥉Brazilian Team
🏆Bonus Prize winner: @ETHBiodivX
Learn more about the teams in our video below. ⬇️
cc: @ColoradoMesaU
We're delighted to announce that our @LimelightRF team made it on the podium of #XPRIZERainforest! Regardless of the result (🥇🥈or🥉???), we're very proud of @traitlab's contribution to @LimelightRF through our innovative canopy tree mapping solution🌳Stay tuned for more!
In this new study from our lab, @myriam_cloutier shows that segmentation of temperate trees from drone imagery works best at the onset of autumn when leaves start of deciduous trees start to change color. authors.elsevier.com/c/1jMU2…