Flagger | Content Author | Content | Reason | Flag Created | Resolved by | Resolution |
---|---|---|---|---|---|---|
mftasp | Tasmanian blue gum (Eucalyptus globulus) |
super-trigger-happy computer vision. Hundreds of obs are incorrect. |
Oct. 23, 2019 02:31:26 +0000 | mftasp |
Most observations checked |
I've started going through and 95%+ of out-of-Australia observations I've looked at are incorrect. Keep in mind there are thousands.
I suspect the west coast of North America will have a higher proportion of correct ones, as will western Europe, areas that have seen commercial plantations of it.
I'd say so far in Asia, 80% were random trees, 10% or so were other myrtaceae, a few remaining were other eucalypts, and a handful were recognisable as E. globulus.
There are a few, other than just having an idea of what the tree looks like from a distance, which you only get from practice.
Eucalyptus globulus has either one of three large buds/flowers/fruit. Never 7. The vast majority of germplasm exported to the world were Tasmanian, 1-fruited provenances.
The bark should be smooth, decorticating, with a stock of rough bark at the base.
The leaves are large, lanceolate and falcate. Adult foliage is dark green and relatively glossy. Juvenile foliage is glaucous (blue).
They should do for a start, and thanks for the offer to help! I've done an easy few hundred in the Asia-Pacific already.
Thanks @thebeachcomber. I've done a few passes and it's looking a fair bit better.
There were particularly a lot of distant, unrecognisable trees getting tagged as E. globulus when they were clearly not, and I wonder whether computer vision was particularly likely to offer up this ID for blurry distant trees.
Are most of the wrong ones outside Australia, or just wrong everywhere in general?