Journal archives for February 2023

February 21, 2023

Identifier Profile: Fly Identifiers

It’s been a while since the last Identifier Profile (sorry about that!), so I thought I’d feature the top five identifiers of syrphid flies (commonly known as hover flies or flower flies, among other names) in North America this month and put together a sort of “oral history” of how they’ve built both resources and a community of identifiers for this taxa and region. I think it’s a great model that could work for other places and taxa around the world.

These identifiers are:

@trinaroberts - Trina Roberts

@edanko - Even Dankowicz 

@zdanko - Zachary Dankowicz 

@catherine_g - Catherine C. Galley 

@upupa-epops - Caleb Scholtens

Collectively, as of February 21st, 2023, they’ve added nearly 300,000 IDs to syrphid observations in North America. 

I’ve tried to make things roughly chronological here, but with an eye on giving it a narrative flow. Some quotes have been lightly edited for clarity. The “Fly Guide” referred to multiple times can be found here.


Even: In high school, I started photographing insects and identifying them with BugGuide. Almost every day I'd find new creatures I'd never seen before.

I first joined iNaturalist for feedback on plants, mollusks, and other groups not covered by BugGuide. I realized that iNaturalist relies on everyone to help, and quickly got started identifying isopods, millipedes, and other groups where a lot of observations were unidentified. In mid-2017, I started going through all the unidentified true flies of “Tropical Asia” (India to Vietnam and New Guinea) in the hopes of finding photos of Cacatuopyga flies (large, striking wasp-mimics only known from a handful of museum specimens). This eventually developed into my current fly specialty.

Caleb: I found out about iNaturalist from a blog post by @mikeburrell in spring 2017, I would've been in my second-last year of high school then. I was already hooked on eBird for keeping track of the birds I'd observed and had been craving something similar for other wildlife, so as soon as I figured out iNat I started uploading old photos of bugs, plants, herps etc. I had always been curious about bugs but had no idea how to learn about them, so posting photos and quickly having people put names to them was magical. 

Wanting to give back (and bored and procrastinating from schoolwork...) I started dabbling with identifying. I had the most experience with birds, but I found that helping with them wasn't very productive because there were so many other people identifying them already. After trying a couple other groups I settled on hover flies because they were relatively easy to identify, and at the time they weren't getting much attention from anyone else.

Even: Flower flies (Syrphidae) are a group of charismatic pollinators that can be fun to identify. Starting in mid-2018, I found it useful to organize checklists and tables of diagnostic characters for different species. Martin Hauser (@phycus), Bill Dean (@billdean), and several others set the stage with their notes on BugGuide. I shared my “notes” with the iNat community, but only a handful of highly-motivated people used them. My “note documents” were too dense to be quickly consulted, even for me!

Caleb: My focus with hover flies has been particularly on Calligraphers (Toxomerus), which are appealing to me because they're abundant and their abdomen patterns are really beautiful, plus they have a cool common name. Those things do apply to a lot of other syrphids though!

Once I had (temporarily) identified all the Toxomerus in Canada and the US, I started observing and identifying other hover flies, particularly Allograpta and other Syrphini, mainly using BugGuide comments as a resource. The new field guide for syrphids of northeastern North America came out a bit later and was helpful as well.

At some point I realized that writing out how to identify them in a journal post would have the double benefit of consolidating what I've learned so I can reference it when I forget something, and teaching other people if they're interested in helping (or just curious about their own observations).

Even: Caleb Scholtens (@upupa-epops) had some experience with Globetails (Sphaerophoria), Aphideaters (Eupeodes), and some other hard syrphid groups with a lot of similar species. I messaged him and he was enthusiastic about working together on guides to these types of flower flies.

I hoped that putting together a “field guide” of sorts could help people and encourage budding identifiers. Caleb and I started by making a “field guide” to common species of the tribe Syrphini in the southwestern USA. The guide was a big hit, and is now linked along with many other recent guides at https://sites.google.com/view/flyguide/species-guides/syrphidae.

Trina: I am a biologist (mammalogist and biogeographer) by training, and I work at the Natural History Museum of L.A. County, but that’s more because I’m a naturalist at heart than the other way around. As of 2019 I had started to use and enjoy iNat frequently as an observer, and I had done some very general IDing for the museum during the City Nature Challenge.

I was posting observations and other people were IDing them, but I wasn’t returning the favor, and I felt it was time I started contributing more to do my share.  So, over the winter I had decided some more IDing would be a good way to do two things at the same time: learn some California taxa, and do my part for the iNat system. My original goal was to ID ten times my number of observations and then decide whether I wanted to keep going.

Catherine: Even’s idea of creating The Fly Guide as an online document specifically designed to help identify live specimens in photographs has been especially helpful. I learned about it in September 2020 and started using it immediately to identify my observations. It is a wonderful resource but for the neophyte I was, there were so many characteristics and subtle nuances between species that at first I could not even notice some of the differences. Using the visual keys was a challenge but I was learning and as I became more familiar with the features I had to look for, I started taking more identifiable photos. I never imagined that I was just at the beginning of an amazing adventure.

Trina: When the pandemic hit and I found myself suddenly spending a lot more time than usual in my apartment, it seemed like a perfect time to pick up the IDing idea.  After some false starts I decided I needed to pick a group to focus on. It had to be something with some interesting diversity, preferably found in my little urban backyard, fun to look at, with enough field marks that a non-specialist could start contributing pretty fast, and with enough of a backlog that identifiers were needed.  And I quickly realized that it also needed to have some active identifiers who didn’t mind being tagged and asked questions, because being able to get some feedback is crucial to learning.  Syrphidae was perfect!  There were a lot of un-IDed and half-IDed observations that included common and distinctive species, and thanks to the existing regular IDers there were friendly and responsive people who would confirm, correct, discuss, and appreciate as needed.

[It became apparent early on] that there are not very many good, thorough, identification resources for photos. When Even launched the first Fly Guide mini-guides that really sped up my learning process for common species that I hadn’t figured out yet, and I still point other observers toward those frequently.

Even: After Caleb and I shared our first flower fly guides, Trina Roberts (@trinaroberts) began making a large number of IDs on west coast syrphid observations. Trina has taught us a lot in the discussions on observations, and she is now the top identifier for Syrphidae with >130,000 identifications.

Zachary: I’m a junior in high school in Bethesda, Maryland. In December of 2020 - winter break of my freshman year - I was incredibly bored and had nothing better to do than to play video games and read books. My brother Even, who was home because of the pandemic, offered to help me find something to do, and at 12:51 p.m. on December 23, 2020, I made my first real identification on iNaturalist, of a Palpada pusilla (Syrphidae) in Florida. 

I continued identifying the Palpada on iNaturalist and built up basic expertise of the Nearctic species. A few weeks later, Even suggested that I could work on creating an illustrated dichotomous key to the species of Palpada in the USA and Canada, and I agreed enthusiastically. We finished the key a few months later and Even posted it on the Fly Guide, where it is now available. Over the next year, Even and I worked together to create illustrated keys for all of the genera of the tribe Eristalini in the Nearctic region.

Caleb: I am really impressed with how much they've covered already and how regularly they make progress on covering new taxa. I helped with the Syrphini guide but most of the work Even did with hover flies was as I was getting busier with university and I couldn't contribute as much as I would've liked to.

Catherine: Aware that the number of observations grew exponentially, Even would mention The Fly Guide and suggest that I become an identifier. It was nice of him but I made sure to let him know that I could not do it because I had no credentials and identifying was scary. I enjoyed iNaturalist and I did not want to fill it with errors, despite Even and later Zachary reiterating that making mistakes is the way to learn.

Zachary: In July of 2021, we thought it might be nice to get some more people interested in syrphids through what Even calls “identifying parties” on Zoom. I decided to ask on the iNaturalist Forum to see if anyone would be interested. And boy, were they!

Catherine: I attended the first class, which took place at the end of August and soon it became a series of weekly sessions that have made a huge difference in helping me learn more and gain confidence. Zachary created excellent videos and the informal meetings have been conducive to ask questions and practice identifying in a fun way.

Trina: I was happy to be able to help review some of the Fly Guide keys, but I haven’t found the time (yet) to contribute any of my own. I have tried to do what I can to support the sense of community, because it is definitely part of what keeps me coming back and I hope I can help make this welcoming and fun for others. The collaboration among identifiers is a big reason why this works; we can fix each other’s mistakes and consult on the tricky cases, and we all learn more in the process.

Catherine: Little by little I started expanding the number of species I could identify and Trina has always been willing to help and explain her reasoning. I try to identify syrphid flies regularly and although I am still making mistakes I know more than I did,  I am still learning, and I am incredibly grateful to Even, Zachary, and Trina for sharing their passion and empowering people like me.

Zachary: After that success, we met, and continue to meet, almost weekly and have developed a regular gang, notably including Catherine Galley, whose studious efforts, stemming from the meetings, have landed her a top 10 spot for global syrphid identifiers.

Caleb: Thankfully several syrphid researchers are active on iNat now (possibly because of me emailing them with questions multiple times?), which is super helpful for making progress outside of North America where there are even fewer resources to go off of (@gilfelipe, @phycus, and @ximo_mengual have been quite helpful). It gets tougher there especially because there are so many undescribed species, and even described species often don't have documentation that's helpful for photo identification.

Even: There are many ways for people to learn more and give back to the community by making identifications. Making IDs just by looking for a basic photo match is a bad idea with flies, but the guides we’ve put together should help if used carefully. The best way to learn quickly is to try and ID a lot of observations, and hold back and ask a lot of questions whenever you aren’t sure! We’re always happy to see new people making IDs.

Catherine: In December 2021, Mike Quinn (@entomike), a coleopterist at the University of Texas at Austin, sent me an invitation to the open house organized by the Texas A&M University Insect Collection. I knew that collections existed but I could not think of what I could do there. I contacted a friend, Scott Longing (@scottlonging) who is professor of entomology at Texas Tech University and asked him if it would make sense for me to go there. Scott suggested that I visit the TTU Invertebrate collections and he put me in touch with Jennifer Giron (@jcgiron) the acting curator.

At the beginning of January 2022, Jennifer gave me a tour, showed me some of the syrphid flies, and let me observe one under the microscope. I was amazed that I could recognize that specimen. She invited me to volunteer, which I accepted. She trained me and one year later I am finishing digitizing and identifying most of the syrphid collection. Furthermore during this year Mike has been instrumental in teaching me how to use the collection in relation to my iNaturalist observations.

Trina: I had no intention of becoming a top identifier in anything, but it turned out to be a lot of fun and I didn’t want to stop when I’d reached my original goal. Each observation is a little puzzle to solve, and there’s always something new to learn.

Caleb: I finished my undergraduate this past spring and am working in environmental consulting now; with all of that I've had less time for identifying recently so it's great to see a lot more people identifying hover flies now! It's crazy how much more accessible knowledge and expertise about nature is now; for example I see young birders picking up skills and knowledge much faster than I could have because of all the resources and more experienced birders they have easy access to online. If you're inspired to create your own guides, don't feel pressure to make them super fancy or cover a ton of species or regions. If you feel like there should be a better ID resource for a group of 3 underrated species in your state, make it! And don't be afraid to ask people with more expertise for questions or review; they're usually happy to share knowledge about the species they're passionate about.

(Top photo: Oblique-banded Pond Fly (Sericomyia chrysotoxoides) by @hill_jasonm (CC-BY-NC). Photos of identifiers from top to bottom: Even Dankowicz, Caleb Scholtens, Catherine C. Galley, Zachary Dankowicz, and Trina Roberts.)


In order to keep this blog post manageable I did only cover the top five syrphid identifiers for North America, but they mentioned that the iNat community at large, as well as the help of experts, were vital to building the resources and community described above. I want to share one last thing from Trina, which I think is really important.

…In a group with this many observations, everyone’s contributions help us keep (sort of) on top of the pile of IDs. As well as giving credit where it’s due, it’s important that people know you don’t have to do tens of thousands of IDs to have it make a difference, and I sometimes worry that the focus on top identifiers and leaderboards can convey that impression! If you look at syrphids broken into subfamilies or genera, or locally within individual states, you’ll see lots of other names on the ID leaderboards, and their effort is critical. @matthewvosper catching up on the backlog for Eristalis and Syrphus (and making the Syrphus key); @coolcrittersyt and others who stay on top of observations for a particular state or region; @jane41 and @spencerchau sorting various taxa; @johnklymko, @phycus, @kevinmoran, @billdean and others being willing to discuss the tricky or less common ones; all the people sorting Diptera and Pterygota and Insects into families to begin with… there are really quite a lot of people involved in one way or another.


- you can check out past meeting invites on the Flies of the US and Canada project’s journal page.

- and/or join the fly identification group email list here.

Posted on February 21, 2023 10:24 PM by tiwane tiwane | 44 comments | Leave a comment

February 23, 2023

A new Computer Vision Model (v2.1) including 1,770 new taxa

We released a new computer vision model today. It has 71,286 taxa, up from 69,966. This new model (v2.1) was trained on data exported last month on January 15th and added 1,770 new taxa.

Why v2.1 and not v1.7? As we mentioned in August, we have been training our new computer vision models using a transfer learning strategy.

All of the models that we have released since August (v1.1 - v1.6), one every month, were trained based on the same source model. We call that source model v1.0. The v1.0 model started training in 2021 on 55,000 taxa, 27 million photos, and trained for about 4 months (80 epochs).

While we've been transfer learning new production models, we've also been working on a new source model. We call this new source model v2.0. The v2.0 model started training in 2022 on 60,000 taxa, 30 million photos, and trained for about 9 months (200 epochs). All of the additional data and training time have produced a better source model, which in turn is making better final production models. The model we released today was the first model based on the v2.0 source model (v2.1). Note from the figure below that v2.0 won't ever be released since was trained on data/a taxonomy that is now over 9 months out of sync, which is why we are releasing v2.1 trained (via transfer learning) on data/a taxonomy from January 15th. Also note that we trained v1.7 as a backup in case v2.1 didn't evaluate well. But since v2.1 performed significantly better, we won't be releasing v1.7 and will continue releasing models derived from the v2.0 base until we move to v3.0 in the next 9 to 12 months.

Thanks to NVIDIA for the generous hardware grant that made all of this training possible!

Taxa differences to previous model

The charts below summarize these 1,465 new taxa using the same groupings we described in past release posts.

By category, most of these 1,465 new taxa were insects and plants

Here are species level examples of new species added for each category:

Click on the links to see these taxa in the Explore page to see these samples rendered as species lists. Remember, to see if a particular species is included in the currently live computer vision model, you can look at the “About” section of its taxon page.

We couldn't do it without you

Thank you to everyone in the iNaturalist community who makes this work possible! Sometimes the computer vision suggestions feel like magic, but it’s truly not possible without people. None of this would work without the millions of people who have shared their observations and the knowledgeable experts who have added identifications.

In addition to adding observations and identifications, here are other ways you can help:

  • Share your Machine Learning knowledge: iNaturalist’s computer vision features wouldn’t be possible without learning from many colleagues in the machine learning community. If you have machine learning expertise, these are two great ways to help:
  • Participate in the annual iNaturalist challenges: Our collaborators Grant Van Horn and Oisin Mac Aodha continue to run machine learning challenges with iNaturalist data as part of the annual Computer Vision and Pattern Recognition conference. By participating you can help us all learn new techniques for improving these models.
  • Start building your own model with the iNaturalist data now: If you can’t wait for the next CVPR conference, thanks to the Amazon Open Data Program you can start downloading iNaturalist data to train your own models now. Please share with us what you’ve learned by contributing to iNaturalist on Github.
  • Donate to iNaturalist: For the rest of us, you can help by donating! Your donations help offset the substantial staff and infrastructure costs associated with training, evaluating, and deploying model updates. Thank you for your support!
Posted on February 23, 2023 06:27 PM by loarie loarie | 32 comments | Leave a comment