Quantifying behavior is crucial for many applications in neuroscience. Videography provides easy methods for the observation and recording of animal behavior in diverse settings, yet extracting particular aspects of a behavior for further analysis can be highly time consuming.
Here we present an efficient method for markerless pose estimation based on transfer learning with deep neural networks that achieves excellent results (i.e. you can match human labeling accuracy) with minimal training data (typically 50-200 frames). We demonstrate the versatility of this framework by tracking various body parts in multiple species across a broad collection of behaviors.
The package is open source, fast, robust, and can be used to compute 3D pose estimates. Please see the original paper and the latest work below.