LifeCLEF 2019 Plant
Image-based plant identification on Amazonian flora
Note: Do not forget to read the Rules section on this page
Automated identification of plants has improved considerably in the last few years. In the scope of LifeCLEF 2017 and 2018 in particular, we measured impressive identification performance over 10K species. However, these 10K species, mostly living in Europe and North America, only represent the tip of the iceberg. The vast majority of the species in the world (~369K species) actually lives in data deficient regions and the performance of state-of-the-art machine learning algorithms on these species is unknown and presumably much lower because of the weak amount of training data. Thus, the main focus of the 2019 edition of PlantCLEF will be to evaluate automated identification on the flora of such data deficient regions.
he goal of the task is return the most likely species for each observation of the test set (an observation being a set of images of the same individual plant and the associated metadata such as date, gps, author). A small part of the observations in the test set will be re-annotated by several experts so as to allow comparing the performance of the evaluated systems with the one of highly skilled experts.
We provide a new dataset of 10K species mainly focused on the Guiana shield and the Amazon rainforest (known to be the largest collection of living plants and animal species in the world). The average number of images per species in that new dataset will be much lower than the dataset used in the previous editions of PlantCLEF (about 10 vs. 100). Many species will contain only a few images and some of them might even contain only 1 image.
The training data is now available (see the “Dataset” tab). The test set to be predicted will be delivered around the 1st of March 2019.
Participants are allowed to use complementary training data (e.g. for pre-training purposes) but at the condition that (i) the experiment is entirely re-produceable, i.e. that the used external resource is clearly referenced and accessible to any other research group in the world, (ii) the use of external training data or not is mentioned for each run, and (iii) the additional resource does not contain any of the test observations.
More practically, the run file to be submitted has to contain as much lines as the number of predictions, each prediction being composed of an ObservationId (the identifier of a specimen that can be itself composed of several images), a ClassId, a Probability and a Rank (used in case of equal probabilities). Each line should have the following format: <ObservationId;ClassId;Probability;Rank>
Here is a short fake run example respecting this format for only 3 observations: fake_run
As soon as the submission is open, you will find a “Create Submission” button on this page (just next to the tabs).
Comparison with PlantCLEF 2018
Before the opening of the PlantCLEF 2019 submissions, we encourage participants to train their system and submit runs to the challenge of last year (dealing with 10K species of data-abundant regions). We therefore re-open a new submission round and leaderboard on the challenge page: ExpertCLEF2018. Note that the best performing models of last year have been shared by CVUT at the following URL: http://ptak.felk.cvut.cz/personal/sulcmila/models/LifeCLEF2018/
The two main evaluation metrics will be the top-1 accuracy on 1) the fraction of the test set identified by the pool of experts, 2) on the whole test set.
Pre-trained plant identification models (from ExpertLifeCLEF 2018)
In order to support research in fine-grained plant classification, CVUT shares the pre-trained Inception-v4 and Inception-ResNet-v2 CNN models from their winning submission to the ExpertLifeCLEF 2018 Plant identification task. The pre-trained models may be a good starting point for the participants to LifeCLEF 2019: http://ptak.felk.cvut.cz/personal/sulcmila/models/LifeCLEF2018/
- Discussion Forum : https://www.crowdai.org/challenges/lifeclef-2019-plant/topics
We strongly encourage you to use the public channels mentioned above for communications between the participants and the organizers. In extreme cases, if there are any queries or comments that you would like to make using a private communication channel, then you can send us an email at lifeclef-org[AT]inria[DOT]fr
You can find additional information on the challenge here: https://www.imageclef.org/PlantCLEF2019
Results (tables and figures)
LifeCLEF 2019 is an evaluation campaign that is being organized as part of the CLEF initiative labs. The campaign offers several research tasks that welcome participation from teams around the world. The results of the campaign appear in the working notes proceedings, published by CEUR Workshop Proceedings (CEUR-WS.org). Selected contributions among the participants, will be invited for publication in the following year in the Springer Lecture Notes in Computer Science (LNCS) together with the annual lab overviews.