diplomat deeplabcut analyze_frames

Summary

Takes a DIPLOMAT Frame Store file (.dlfs) and makes predictions for the stored frames, using whatever predictor plugin is selected. This allows for the video to be run through the Deep Neural Network once, and then run through several prediction algorithms as many times as desired, saving time. It also allows for frames to be processed on one computer to be transferred to another computer for post-processing and predictions.

Usage

diplomat deeplabcut analyze_frames [-h] --config FILE --frame_stores
                                          [FILE, ...]|FILE [--predictor STR]
                                          [--save_as_csv BOOL]
                                          [--multi_output_format 'default'|'separate-bodyparts']
                                          [--video_folders [FILE, ...]|FILE]
                                          [--num_outputs INT] [--shuffle INT]
                                          [--training_set_index INT]
                                          [--predictor_settings {STR: VAL, ...}]

Options

-h, --help            show this help message and exit
--config FILE, -c FILE
                      The path to the DLC config to use to interpret this
                      data. The .DLFS will inherit the neural network of
                      this project, allowing for frame labeling using this
                      project.
--frame_stores [FILE, ...]|FILE, -fs [FILE, ...]|FILE
                      The paths to the frame stores (.dlfs files), string or
                      list of strings.
--predictor STR, -p STR
                      A String, the name of the predictor plugin to be used
                      to make predictions. If not specified, defaults to the
                      segmented frame pass engine
                      ("SegmentedFramePassEngine").
--save_as_csv BOOL, -sac BOOL
                      A Boolean, True to save the results to the human
                      readable .csv format, otherwise false.
--multi_output_format 'default'|'separate-bodyparts', -mof 'default'|'separate-bodyparts'
                      A string. Determines the multi output format used.
                      "default" uses the default format, while "separate-
                      bodyparts" separates the multi output predictions such
                      that each is its own body part.
--video_folders [FILE, ...]|FILE, -vf [FILE, ...]|FILE
                      None, a string, or a list of strings, folders to
                      search through to find videos which correlate to the
                      .dlfs files. If set to None, this method will search
                      for the corresponding videos in the directory each
                      .dlfs file is contained in.
--num_outputs INT, -no INT
                      int, default
--shuffle INT, -s INT
                      int, optional. An integer specifying the shuffle index
                      of the training dataset used for training the network.
                      The default is 1.
--training_set_index INT, -tsi INT
                      int, optional. Integer specifying which
                      TrainingsetFraction to use. By default the first (note
                      that TrainingFraction is a list in config.yaml).
--predictor_settings {STR: VAL, ...}, -ps {STR: VAL, ...}
                      Optional dictionary of strings to any. This will
                      specify what settings a predictor should use,
                      completely ignoring any settings specified in the
                      config.yaml. Default value is None, which tells this
                      method to use the settings specified in the
                      config.yaml.