Introducing the iNaturalist 2018 Challenge

Posted by Yang Song, Staff Software Engineer and Serge Belongie, Visiting Faculty, Google Research


Thanks to recent advances in deep learning, the visual recognition abilities of machines have improved dramatically, permitting the practical application of computer vision to tasks ranging from pedestrian detection for self-driving cars to expression recognition in virtual reality. One area that remains challenging for computers, however, is fine-grained and instance-level recognition. Earlier this month, we posted an instance-level landmark recognition challenge for identifying individual landmarks. Here we focus on fine-grained visual recognition, which is to distinguish species of animals and plants, car and motorcycle models, architectural styles, etc. For computers, discriminating fine-grained categories is challenging because many categories have relatively few training examples (i.e., the long tail problem), the examples that do exist often lack authoritative training labels, and there is variability in illumination, viewing angle and object occlusion.

To help confront these hurdles, we are excited to announce the 2018 iNaturalist Challenge (iNat-2018), a species classification competition offered in partnership with iNaturalist and Visipedia(short for Visual Encyclopedia), a project for which Caltech and Cornell Tech received a Google Focused Research Award. This is a flagship challenge for the 5th International Workshop on Fine Grained Visual Categorization (FGVC5) at CVPR 2018. Building upon the first iNaturalist challenge, iNat-2017, iNat-2018 spans over 8000 categories of plants, animals, and fungi, with a total of more than 450,000 training images. We invite participants to enter the competition on Kaggle, with final submissions due in early June. Training data, annotations, and links to pretrained models can be found on our GitHub repo.

iNaturalist has emerged as a world leader for citizen scientists to share observations of species and connect with nature since its founding in 2008. It hosts research-grade photos and annotations submitted by a thriving, engaged community of users. Consider the following photo from iNaturalist:

The map on the right shows where the photo was taken. Image credit: Serge Belongie.

You may notice that the photo on the left contains a turtle. But did you also know this is a Trachemys scripta, common name “Pond Slider?” If you knew the latter, you possess knowledge of fine-grained or subordinate categories.

In contrast to other image classification datasets such as ImageNet, the dataset in the iNaturalist challenge exhibits a long-tailed distribution, with many species having relatively few images. It is important to enable machine learning models to handle categories in the long-tail, as the natural world is heavily imbalanced – some species are more abundant and easier to photograph than others. The iNaturalist challenge will encourage progress because the training distribution of iNat-2018 has an even longer tail than iNat-2017.

Distribution of training images per species for iNat-2017 and iNat-2018, plotted on a log-linear scale, illustrating the long-tail behavior typical of fine-grained classification problems. Image Credit: Grant Van Horn and Oisin Mac Aodha.

Along with iNat-2018, FGVC5 will also host the iMaterialist 2018 challenge (including a furniture categorization challenge and a fashion attributes challenge for product images) and a set of “FGVCx” challenges representing smaller scale – but still significant – challenges, featuring content such as food and modern art.

FGVC5 will be showcased on the main stage at CVPR 2018, thereby ensuring broad exposure for the top performing teams. This project will advance the state-of-the-art in automatic image classification for real world, fine-grained categories, with heavy class imbalances, and large numbers of classes. We cordially invite you to participate in these competitions and help move the field forward!


Acknowledgements

We’d like to thank our colleagues and friends at iNaturalistVisipedia, and FGVC5 for working together to advance this important area. At Google we would like to thank Hartwig Adam, Weijun Wang, Nathan Frey, Andrew Howard, Alessandro Fin, Yuning Chai, Xiao Zhang, Jack Sim, Yuan Li, Grant Van Horn, Yin Cui, Chen Sun, Yanan Qian, Grace Vesom, Tanya Birch, Wendy Kan, and Maggie Demkin.

9 Comments

  • I and also my friends appeared to be examining the excellent things located on your site while suddenly developed a horrible suspicion I had not thanked you for those techniques. The men came as a consequence happy to see them and have now truly been tapping into them. We appreciate you indeed being well considerate and then for pick out certain smart issues most people are really needing to learn about. My personal honest apologies for not expressing gratitude to you sooner.

  • yeezy boost says:

    I in addition to my friends have been examining the good suggestions located on your site and all of a sudden got a terrible feeling I had not expressed respect to the web site owner for those secrets. Most of the boys happened to be consequently happy to learn all of them and have in effect definitely been having fun with them. We appreciate you turning out to be so thoughtful and also for opting for some good themes millions of individuals are really desirous to be aware of. My very own sincere regret for not expressing gratitude to earlier.

  • Thank you so much for giving everyone an exceptionally wonderful opportunity to read in detail from this web site. It really is very sweet plus full of a good time for me personally and my office peers to visit your site at minimum 3 times in a week to see the latest guides you will have. And definitely, I am also actually fascinated with the eye-popping opinions you serve. Some 4 ideas on this page are really the best I’ve had.

  • Thank you so much for giving everyone an extraordinarily remarkable chance to read in detail from here. It’s always very lovely and also packed with a lot of fun for me personally and my office friends to search your website at the least three times in one week to see the fresh issues you have. And lastly, I’m just actually contented with your dazzling solutions served by you. Some 3 areas on this page are really the most efficient we have all had.

  • I precisely wished to appreciate you all over again. I am not sure the things I might have used without the type of tactics revealed by you about such theme. Certainly was an absolute scary setting in my opinion, but finding out your well-written fashion you resolved that took me to leap over happiness. I am just thankful for the guidance and in addition have high hopes you are aware of a powerful job you’re undertaking educating others all through your site. I know that you’ve never encountered any of us.

  • I want to voice my love for your kindness for folks who really need help with this particular question. Your personal commitment to passing the message all-around appeared to be wonderfully practical and has always encouraged some individuals just like me to get to their objectives. Your amazing insightful help and advice means much a person like me and additionally to my colleagues. Thank you; from everyone of us.

  • I would like to point out my love for your kind-heartedness for folks who absolutely need help on this field. Your real dedication to getting the solution across has been exceptionally helpful and have surely made somebody like me to arrive at their desired goals. The valuable help signifies much a person like me and much more to my fellow workers. Best wishes; from all of us.

  • adidas nmd says:

    I wanted to post you this tiny word to be able to give many thanks as before just for the wonderful views you’ve provided on this website. It is so remarkably open-handed with people like you to give publicly exactly what a lot of folks could possibly have distributed for an e-book to help make some money on their own, most notably considering that you might have tried it in the event you wanted. Those inspiring ideas likewise worked to be a good way to be aware that other individuals have the same interest similar to mine to know the truth a whole lot more concerning this problem. I’m sure there are many more enjoyable moments ahead for many who scan through your website.

  • I wish to show some appreciation to this writer for bailing me out of this particular predicament. As a result of surfing around through the the net and obtaining basics which are not productive, I was thinking my life was done. Existing minus the answers to the issues you have sorted out by way of your posting is a serious case, as well as the kind which could have adversely damaged my career if I had not come across your website. Your main understanding and kindness in controlling a lot of things was vital. I don’t know what I would have done if I had not discovered such a step like this. I can now relish my future. Thanks for your time so much for this skilled and effective guide. I won’t be reluctant to suggest the website to anyone who ought to have guidance on this area.

Leave a Reply