SLEAP project for pose estimation of naive and chonicially implanted bearded dragon lizards
videos were acquired at 4k. cropped and downsampled to 720 with NVENC H264 encoding.
Latest model trained on both implanted and non implanted animals. with error of ~3 pixels.
implant with color tape - red and green. but no longer used for for tracking.
usage: Model = "path\to\models\modelFile.slp"
BatchSize = 512
VideoIndex = 0
%track sleap-track -m Model --gpu 0 --batch_size BatchSize --video.index VideoIndex --verbosity rich
%to save predictions as .h5: %#predictions.slp file PredictionsFile= "H:\scratch\GLL1595_videos\predictions\cam0_2023-09_25_29_10-12_14_116_IRavi_labeled_241023_720_EM.slp.231031_063130.predictions.slp"
sleap-convert PredictionsFile --format analysis
%to save predictions as csv
sleap-convert PredictionsFile --format analysis.csv