• data

  includes all the data files (.txt) and the drawings (.png) of the participants.

• ref_pics

  includes all the reference pictures.

• ref_pics_crop

  includes all the cropped reference pictures (with only the active pixels).

• sketch_simplification

  includes files for implements sketch simplification.

• articles

  includes relevants artical for our project.

• concat

  includes results of the Pix2Pix network (input, label, prediction) as one picture.

• dataset

  includes results of the Pix2Pix network (input, label, prediction) separte.

• participant_output_data

  includes participant draws and graphs.

• pickle

  includes pickle data.

• clustered_draws

  Tal?

This is the atomic unit it our data. Stroke is holding information of the time, (x,y) location and pressure of the stroke. Every one of them represent as np.array (each cell represent a different sampling in the current stroke).

• get_feature(feature_name): return the np.array that represent the given feature name.
• average(feature_array): return the average of the given feature array.
• length(): return the total geometric length of the stroke.
• time(): return the total time of the stroke.
• is_pause(): return true iff the stroke is "Pause stroke"
  • There are two type of Stroke: "Normal stroke", and "Pause stroke". "Pause stroke" is create to fill the empty space left by the Pause, so actually it's not a real data. By default, Stroke is define as 'normal' stroke, for creating 'Pause stroke' you need to add pause=True in the constructor call. The 'Pause stroke' is taking the total time of the pause, and divide it into pieces, with time stamps of 0.017 seconds. The pressure is define as 0, and all other fields (excepts 'time') is define as -1.

Represent a drawing. it holding list of Strokes, which together represent a complete drawing. It also holding a path to the reference picture (what the participant was trying to draw), and a path to the actual picture that he draw. For creating a Drawing, you need to call: Analyzer.create_drawing(path-to-drawing.txt)

• total_length(): return list of size 2, the first value is the total geometric length of the drawing. The second value is array of all the length of each stroke (real stroke, not pause) in the drawing.
• speed_vs_time(), pressure_vs_time(), length_vs_time() : In these functions, we plot graph of the feature (speed, pressure, length) as function of time. In speed_vs_time and pressure_vs_time, every point in the graph is represent a stroke: x value is the starting time of the stroke, and the y value is the average of the feature, in the stroke. In length_vs_time, the x value is the same, but y value is the total geometric length of the stroke. You can choose if you want to see the pause-strokes in the graph, or not. The default is to plot the graph without them, and if you want to see them, just call the functions with: pause=True in the parameters.
• plot_picture(): In progress. Should plot the reference picture and the actual drawing, in the same resolution.
• plot_crop_image(): plot the picture with only the active pixels area.
• group_strokes(euc_dist_threshold, dist_threshold, ang_threshold): This function dividing the draw into groups of strokes which create a clutstes for the draw. Example for clustering:
  • 

Represent a participant. The constructor get the name of the participant (for example: 'aliza') and it holding a list of Drawing of all the drawing of the participant.

• plot_participant_pictures(): plot all the drawings of the participant (by using plot_picture() of Drawing class).
• export all participant draws and graphs into "participants_output_data" folder. calculate general means of features and write it to .txt in the same folder.
• simplify_all_clusters(euc_dist_threshold, dist_threshold, ang_threshold): Simplify all the clusters of the participant (which include all the draws).
• searching_match_on_person(p1, load_dict, euc_dist_threshold, dist_threshold, ang_threshold): Find the best match for 2D array p1 in the participant dict
• create_dict(euc_dist_threshold, dist_threshold, ang_threshold): Creating a dict of simplify cluster and style cluster of the participant. illustration of creating the dict:
  • 

This class including static methods for analyzing the data.

This file include the function find_nearest_neighbor(p1, neighbors), that getting array of 2D points (p1), and list of others arrays points 2D (neighbors), and find the closest array, according to our distance metric.

This file is using for simplify clusters. Cluster is define to be a group of Strokes, so this file getting a cluster, and make it more "simple".

For applying style of on given drawing path, use the following commands:

path = "path-to-input.txt"
draw = Analyzer.create_drawing(path, orig_data=False)
new_draw = transfer_style(draw, <participant_name>, load_person=False, load_dict=False, already_simplified=True)
new_draw.plot_picture(show_clusters=False)

For second run, you can use load_person=True, load_dict=True