Onwards to 2021 ... some major updates!
We have been incredibly fortunate to have funding and support from the Chan Zuckerberg Initiative through the Essential Open Source Software for Science program in 2020 … and we are beyond thrilled to announce, we have been awarded a new round of funding for 2021!
*read the official press release from CZI here (and their medium post!)
images from: https://chanzuckerberg.com/newsroom/czi-awards-4-7-million-for-open-source-software-and-organizations-advancing-open-science/
What we did in 2020:
2020 was an exciting year for #teamDLC.
Firstly, with CZI funding we were able to hire Dr. Jessy Lauer to join us as a lead programer! Jessy joined us at Harvard in January, and we all hit the ground running - our first CZI-DLC user workshop, which brought users from across the world to the campus for a 2.5 day immerse hands-on workshop. Jessy and co worked on a new test suite and continuous integration for DeepLabCut, and joined us on ImageForum SC to assist users.
We also had the amazing opportunity to join for the CZI EOSS meeting (the last trip of 2020, what a highlight though!) See the slides I presented here!
Then in early March we held the first of two planned hackathons — it was amazing. Everyone hacked, shared, laughed, and discussed the joys and struggles of open source science. It was also, of course, the last in-person event for many of us, and sadly will remain the only hackathon of 2020 due to COVID19. But, it also resulted in a new DLC-live! package being integrated into Bonsai and AutoPilot, plus a paper (Kane et al. 2020 bioRxiv, under revision at eLife):
We also over-hauled our documentation, added tens of new “how-to” YouTube videos to our channel, and worked on community guidelines! Essential housekeeping for the growing DeepLabCut consortium and user base!
Mid-year we all moved to EPFL, where the DLC magic continues in the Mathis Laboratory & Mathis Group.
Jessy alongside Alexander and myself have also been working on a major code base update, multi-animal DeepLabCut, which we have under beta release, but the full release will be out within 2020! (there are some exciting updates to come, watch out!)
Towards our efforts on education, we also wrote a Primer on using deep learning for motion capture (pose estimation). We hope this serves as a solid guide for anyone jumping into the field, or even advanced users, for building intuition on how the tools work!
Plus, given we could not hold in-person events after March 2020, we launched an online, self paced DLC course!! http://DLCcourse.deeplabcut.org
We started off 2020 with ~50K downloads of DeepLabCut. Last week we hit 165,000 — THANK YOU to our amazing users! Without the users and the community, this would not be what it is today.
We now have 590 citations to the set of DLC papers, over 49 code contributors, 890 forks, and 1.8K stars—plus 5.2K twitter followers and 700 subscribers on YouTube! Of course many of these metrics have major limitations, but we do enjoy knowing there is a active and growing user base ... looking forward to 2021!
What’s on deck for 2021:
With this new round of funding, we are delighted to keep Jessy Lauer on as a lead-developer for 2021!
Here is what we aim to continue do:
Provide code maintenance, provide code updates to integrate state-of-the-art deep learning advances, and provide user help to code questions.
Some example code base updates: complete the migration from tensorflow 1 to tensorflow 2+
Continue to update documentation, testing suite updates, and further flush out the API.
Assistance on Image Forum SC, Gitter, & GitHub.
New developments:
A new DLC cook-book! This will be a dedicated place for worked code examples, allowing both novice and advanced users to understand and use on more of DLC’s capabilities! With many hidden gems and use cases lurking in the code base, we aim to flush them out through code and data examples!
DLC-Napari: Integrate the DLC stand-alone labeling tools, frame editing, tracking tools, and video editing with napari. That’s right! We are migrating to the latest Python viewer of choice, napari!
