The next sentence is from the example:
The PlantCLEF2022 test dataset contains 26868 plant observations related to 55306 images, meaning that many plants are related to several images.
What does plant observation mean?
For me, I tried to classify all 55306 images into classes in the training dataset.
So, the image named 10251.jpg is classified.
But the submission is suggested to be wrong.
Could you please help me? Thank you.
I got now.
One plant may have several images for different paper, such as flower and leaf.
These info can be looked at https://lab.plantnet.org/LifeCLEF/PlantCLEF2022/test/PlantCLEF2022_test_metadata.csv.
Yes indeed, this is a notable difference with classical image classification problems where we usually compute the performance of a model by averaging its prediction scores on test images. Here the test samples are observations and not images. If an observation is associated with several images, then the predictions must be combined. This also reflects the use of botanists: to ensure the species identification of one single plant, a botanist must often observe several organs (a flower, a fruit, a leaf), even a same organ from different angles (a flower in front view, the same flower in bottom view, side view, etc). A single image can possibly help to identify the genus of a plant, but hardly the species.