Tree species identification ­čî▓­čî│and geolocation­čÄ» are achievable with #DeepLearning object detection and #RGB imagery: doi.org/10.3390/rs1505ÔÇŽ #CNN #forest #biodiversity

>10,000 tree species annotations for deep learning are available online. #RGB orthophotos

@BioMirela Wow so nice, I will read this paper today... Do you use Yolo?

@AlanLacerd Not in this paper, but yes, I have also tried Yolo and SSD.

@BioMirela Awesome paper so far! I have the problem that trees often stand very dense to each other in Schleswig-Holstein, almost not separable by human eye. And I have only 20cm aerial pictures, but RGBI. Would you still try bounding boxes or go for segmentation?

@GattingerTim Hi Tim, this method was applied in mixed forests, so it should work at least for the same species. Well, it depends on what you need. If you need individual tree species, it would still be interesting to try it. Otherwise, segmentation should work well.