Classification will be executed on the base of spectral or spectrally defined features, such as density, texture etc in the feature space.
Step1 . Select features to disciminate between yet-to-be-assessed tropical rain forest plant species and IUCN Red List.
Step2 : Sampling of Training data : Training data may the images you have collected for your training . it should be sampled in order to determine appropriate decision rules.
Step 3. Classification techniques such as supervised or unsupervised learning will then be selected on the basisi of the traning data sets.
I agree with all that Simon says, but would add that it is often extremely difficult to get enough data to do a formal IUCN assessment for tropical forest trees, which is why so few - I'd guess less than 1-2% - have been evaluated. The IUCN document makes clear that it is better to classify a species on the basis of the best available information than to not classify it at all, but in practice people seem to be very reluctant to do this, particularly for plants. For canopy trees, Asma's suggestion of remote sensing data is an attractive idea, but although Greg Asner's group has shown that identifying trees to species may be possible, I have never seen it used in a conservation assessment. I'd guess that if a species was very distinctive in some way that was detectable from space this would be possible.
IUCN Red List office offers on-site and online courses to help people do species assessments. I attended that training some years ago and I have been listing species in Nigeria and parts of West Africa on IUCN red list. Below is a weblink of one I listed some years back: http://www.iucnredlist.org/details/71022951/0
And I am currently working with some tropical ecologists in Europe to list many more tropical plants. Kindly contact me with information to see if the species you are interested in is on the list of species to be listed: [email protected]