DeepLabCut, the software which revolutionizes pattern recognition of movements across behaviors and species has been successfully implemented at INCIA for a year, and is fully operational! DeepLabCut (Mathis et al, Nature Neuroscience 2018) is a very efficient method for markerless pose estimation of body parts based on transfer learning with deep neural networks.
If you are interested in behavioral analyses and/or are already using traditional tracking systems, especially those based on markers (may be impossible to place on very small or aquatics animals) or those based on heuristic methods (such as object segmentation) with poor success: come and try deeplabcut !
Advantages: video processing can be done on your own computer upstream (deeplabcut is freely available at mousemotorlab.org/deeplabcut) and the training process on a dedicated powerful workstation on IMAGYS* facility. The software can be used with minimal to no coding experience.
contact gilles.courtand@u-bordeaux.fr
*IMAGYS is your microscopy and image engineering facility at the INCIA
