Learning to use DeepLabCut
Mackenzie Mathis
Created on November 17, 2020
flow diagram on learning DLC
More creations to inspire you
EDTECH CAFE FALL 2019
Personalized
BIOETHICS
Personalized
HOW TO MAKE AN INFOGRAPHIC
Personalized
BLOODY BRIDE
Personalized
PRACTICE SPEAKING ENGLISH CHOICE BOARD
Personalized
CHOICE BOARD: TO KILL A MOCKINGBIRD
Personalized
DAILY SPECIALS MENU HORIZONTAL INFOGRAPHIC
Personalized
Transcript
Learn to use DeepLabCut!
getting started
Welcome to the DLC Course! Ready to get started? Here, we will walk you through the main steps to get DLC up and running in your hands in no time! You can start at the top, or jump to any topic you might want to learn more about!
#teamDLC
More information
Got Poses? Now what?
Scaling up your analysis pipeline
Step 3: you've got a network, now let's evaulate it's performance
A deeper dive into neural network selection
Step 2: using DeepLabCut on your own data
Learning more about DeepLabCut!
Jumping deeper into Python: resources, more links, and #proTips
Step 1: getting set up with python, anaconda, and your dlc computing environment
How do the neural networks work? How do I best pick a network for my use case? DeepLabCut has several modified neural networks for you to choose from. The powerful ResNet-50, -101, the ultra fast MobileNetV2s, and now EfficientNets, bringing you the latest high performance. Read our paper on comparing these networks: https://arxiv.org/abs/1909.11229 And #DLCProTips on what you might want to use: READ MORE HERE
Contact Developers GitHub / Twitter / YouTube
There are many new packages that take in the outputs of DLC for you to do further analysis!
You can automate your pipeline Using tools like cron jobs, batch processing, DataJoint, AWS, and more, we point you to the tools that can help make your life even easier ....
Your network is only as good as your input data, and of course how you load that data and train it. We provide several tools to evaluate your network, so you can carry on with confidence in your video analysis.
Typically the first step is to locally install DeepLabCut on your computer You can install just for CPU use (i.e., for labeling data), or if you have your own GPU, you can install for this as well. Otherwise, you can use free cloud services, such as google colab! Click away to read more ....
Looking for a deeper dive into Python? There are a lot of amazing resources, jump in here!
We have 4 key papers that help you learn how it works, and how best to use DLC! Mathis et al, Nature Neuroscience 2018 or free link: rdcu.be/4Rep Nath*, Mathis* et al, Nature Protocols 2019 or free link: https://rdcu.be/bHpHN Mackenzie W. Mathis & Alexander Mathis. Deep learning tools for the measurement of animal behavior in neuroscience Current Opinion in Neurobiology Volume 60, February 2020, Pages 1-11 Alexander Mathis, Steffen Schneider, Jessy Lauer, Mackenzie W Mathis - A Primer on Motion Capture with Deep Learning: Principles, Pitfalls, and Perspectives Neuron 2020. https://doi.org/10.1016/j.neuron.2020.09.017
Need help with your own project? There is an active user forum: https://forum.image.sc/tag/deeplabcut
There are many ways to use DeepLabCut: through a full GUI (NO programming required), via. Jupyter Notebooks, or in the command line interface! We have created video tutorials for each of these options: https://www.youtube.com/channel/UC2HEbWpC_1v6i9RnDMy-dfA DeepLabCutEnjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on...YouTube
We post new code release announcements on Twitter (please follow!), and well as #DLCProTips! Search for more details...
Did you know, we have several review papers on deep learning for animal behavior! Some cover the current state of the field, while others are deep dives into the technology, and how best to use them!
You should check out our new paper, which really takes you into important points on neural network selection, data augmentation, and pitfalls to avoid! https://www.cell.com/neuron/fulltext/S0896-6273(20)30717-0 open source: https://arxiv.org/abs/2009.00564
We have a code and pointers to other packages that will help you analyze the output of DLC! https://github.com/DeepLabCut/DLCutils
If you want to make your own customize analysis pipeline, check out: https://scikit-learn.org/stable