Animal behavior analysis tool "DeepLabCut" now supports TensorFlow 2

This is the engineering department of Tegara Corporation

This time, recently our PC production service TEGSYSI would like to introduce information about the animal behavior analysis tool "DeepLabCut", which has been increasingly contacted as software to be used when making inquiries to.

What is DeepLabCut?

Software for animal behavior analysis developed in open source by a team led by Professor Mackenzie Mathis of the Swiss Federal Institute of Technology Lausanne.

By applying the deep learning method, it is a major feature that advanced behavior analysis based on videos can be performed by GUI operation. At present, attention is being paid to the convenience of not requiring complicated command operations by being able to perform GUI operations, and the fact that the functions themselves are extremely useful in animal behavior analysis.

DeepLabCut official website
http://www.mackenziemathislab.org/deeplabcut

 

Support for TensorFlow 2

As mentioned above, DeepLabCut is widely recognized for its usefulness, but until recently it had one technical problem. DeepLabCut is implemented by using TensorFlow as the basis of DeepLearning, but it was based on TensorFlow 1 from the beginning of the release, and support for TensorFlow 2 released after that was not advanced.As a result, it was not possible to support Ampere generation NVIDIA GPUs as it is, and it has been difficult to utilize the latest GPU resources since the latter half of last year.

However, with the release of 2021rc6 in June 2.2, an update that includes support for the long-awaited (?) TensorFlow 3 has made it possible to use Ampere generation NVIDIA GPUs as standard software.

Tensorflow 2.x will be the new default in DeepLabCut!
http://www.mackenziemathislab.org/deeplabcutblog/2021/6/30/tensorflow-2x-is-the-new-default-in-deeplabcut

We have also verified the operation immediately and confirmed that high-speed processing by GPU is possible with DeepLabCut on Windows 10, Ubuntu 18.04, 20.04 equipped with Ampere generation NVIDIA GPU.

[Reference] PC that meets DeepLabCut installation & operation requirements

At Tegara, we sell computers with DeepLabCut installed at our company in response to customer requests.For example, we have a track record of proposing the following machines.

 

We will also propose a specification configuration that meets the customer's request, so we would appreciate it if you could contact us from the TEGSYS website when ordering.

Person who wrote this article : Engineering Department Wada
The date this article was written : 2021.8.10