The gap between the current capabilities of image-based methods for automatic plant
disease identification and the real-world needs is still wide. Although advances have been
made on the subject, most methods are still not robust enough to deal with a wide variety
of diseases and plant species. This paper proposes a method for disease identification,
based on colour transformations, colour histograms and a pairwise-based classification
system. Its performance was tested using a large database containing images of symptoms
belonging to 82 different biotic and abiotic stresses, affecting the leaves of 12 different
plant species. The wide variety of images used in the tests made it possible to carry out an
in-depth investigation about the main advantages and limitations of the proposed algo-
rithm. A comparison with other algorithms is also presented, and some possible solutions
for the main challenges that still prevent this kind of tool to be adopted in practice.