![]() ![]() Of note, DeepLabCut can OUTPERFORM humans. We will look at several aspects in particular: Positioning the camera(s) using 3D printed and off the shelf parts, manual focusing with novel lenses, capturing video in Windows (please consider Ubuntu, you will be happy you did!), using this camera in Windows in Matlab (note, it works natively in Ubuntu Matlab!), capturing time stamped images in python (on Ubuntu), use in LabView, and finally syncing the cameras together so that they capture simultaneously (it’s like reverse Hebbian learning: Those that wire together fire together!!!). ![]() Here we will look at the use of this camera, with special consideration for DeepLabCut (also see ONE Core notes on installation). At the time of writing, this camera is around $7. This little powerhouse can probably get at your needs. We show how to get time stamped frames at 640x480 resolution at greater than 90 frames per second. See Fig 3 of On the inference speed and video-compression robustness of DeepLabCut.Įnter the ever-hackable PS3Eye Camera, henceforth called PS3E (yeah, I’m too lazy to type the ‘ye’). In fact, because of limitations in computation times and data storage, most videos taken with high resolution/speed cameras are immediately down-sampled. ![]() But oftentimes, high resolution at high frame rates is not necessary. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |