Mining Field Data for Tree Species Recognition at Scale
This addresses the challenge of limited expert-labeled data for tree species recognition in forestry, though it is incremental as it builds on existing detection models.
The authors tackled the problem of acquiring individual tree species labels by automatically mining them from public forest inventory data using pretrained tree detection models, achieving close to zero human involvement and showing a beneficial effect from adding noisy or unlabeled data for large-scale species mapping.
Individual tree species labels are particularly hard to acquire due to the expert knowledge needed and the limitations of photointerpretation. Here, we present a methodology to automatically mine species labels from public forest inventory data, using available pretrained tree detection models. We identify tree instances in aerial imagery and match them with field data with close to zero human involvement. We conduct a series of experiments on the resulting dataset, and show a beneficial effect when adding noisy or even unlabeled data points, highlighting a strong potential for large-scale individual species mapping.