ROC/DET plotter following the ISO/IEC 19795-1:2006(E) standard
Parameters allow users to change the configuration of an plotter when scheduling an experiment
| Name | Description | Type | Default |
|---|---|---|---|
| xlabel | The label of the X-axis (horizontal) | string | False Positives (False Match Rate), in % |
| ylabel | The label of the Y-axis (vertical) | string | True Positives (1 - False Non-Match Rate), in % |
| xlim-left | float64 | 0.0 | |
| xlim-right | float64 | 100.0 | |
| ylim-bottom | float64 | 0.0 | |
| ylim-top | float64 | 100.0 | |
| title | The title for this plot | string | ISO/IEC 19795-1:2006(E) ROC |
| title-fontsize | Controls the title font size | uint16 | 10 |
| xaxis_multiplier | The multiplication factor for the X-axis (horizontal) | float64 | 100.0 |
| yaxis_multiplier | The multiplication factor for the Y-axis (vertical) | float64 | 100.0 |
| axis-fontsize | Controls the axis font size (labels and values) | uint16 | 10 |
| legend | Short description of the data, to be added to the plot | string | |
| legend-fontsize | Controls the font size of the legend | uint16 | 12 |
| legend-loc | The location of the legend | string | best |
| grid | If we should draw grid lines or not for the plot | bool | True |
| xaxis_log | If X-axis (horizontal) should be in log-scale | bool | False |
| yaxis_log | If Y-axis (vertical) should be in log-scale | bool | False |
| dpi | Dots-per-inch in raster image formats | uint16 | 60 |
| width | Width of the resulting image in pixels | uint16 | 400 |
| height | Height of the resulting image in pixels | uint16 | 300 |
| content_type | The type of image returned | string | image/png |
| line_attributes | Scatter/Line attributes passed directly to Matplotlib | string | |
| det | If set, plot a DET curve instead of a ROC | bool | False |
xxxxxxxxxx################################################################################ ## Copyright (c) 2016 Idiap Research Institute, http://www.idiap.ch/ ## Contact: beat.support@idiap.ch ## ## This file is part of the beat.examples module of the BEAT platform. ## ## Commercial License Usage ## Licensees holding valid commercial BEAT licenses may use this file in ## accordance with the terms contained in a written agreement between you ## and Idiap. For further information contact tto@idiap.ch ## ## Alternatively, this file may be used under the terms of the GNU Affero ## Public License version 3 as published by the Free Software and appearing ## in the file LICENSE.AGPL included in the packaging of this file. ## The BEAT platform is distributed in the hope that it will be useful, but ## WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY ## or FITNESS FOR A PARTICULAR PURPOSE. ## ## You should have received a copy of the GNU Affero Public License along ## with the BEAT platform. If not, see http://www.gnu.org/licenses/. ## ################################################################################# Makes sure we won't require an X11 connectionimport matplotlibmatplotlib.use('Agg')import numpyimport itertoolsclass Plotter(baselib.Plotter): def setup(self, parameters): super(Plotter, self).setup(parameters) self.line_attributes = parameters.get('line_attributes', '').split('&') self.loc = parameters.get('legend-loc', '') self.title_fontsize = parameters.get('title-fontsize', '') self.xlim_left = parameters.get('xlim-left', '') self.xlim_right = parameters.get('xlim-right', '') self.ylim_bottom = parameters.get('ylim-bottom', '') self.ylim_top = parameters.get('ylim-top', '') self.axis_fontsize = parameters.get('axis-fontsize', '') self.legend_fontsize = parameters.get('legend-fontsize', '') self.line_attributes = [k for k in self.line_attributes if k] self.det = parameters.get('det', False) if self.det: self.ylabel = "False Negatives (False Non-Match Rate), in %" # These are the "Tableau 20" colors as RGB. self.tableau20 = [(31,119,180), (152,223,138), (140,86,75), (199,199,199), (174,199,232), (214,39,40), (196,156,148), (188,189,34), (255,127,14), (255,152,150), (227,119,194), (219,219,141), (255,187,120), (148,103,189), (247,182,210), (23,190,207), (44,160,44), (197,176,213), (127,127,127), (158,218,229)] # Scale the RGB values to the [0, 1] range, which is the format matplotlib accepts. for i in range(len(self.tableau20)): r, g, b = self.tableau20[i] self.tableau20[i] = (r / 255., g / 255., b / 255.) return True def process(self, inputs): fig, ax = super(Plotter, self).prepare_canvas() super(Plotter, self).apply_parameters(ax) ax.set_title(self.title, fontdict={'fontsize':self.title_fontsize}) ax.set_xlabel(self.xlabel, fontdict={'fontsize':self.axis_fontsize}) ax.set_ylabel(self.ylabel, fontdict={'fontsize':self.axis_fontsize}) ax.set_xlim((self.xlim_left,self.xlim_right)) ax.set_ylim((self.ylim_bottom,self.ylim_top)) label = "" kwargs = {} single_experiment = len(inputs) == 1 number_of_positives = [] number_of_negatives = [] n_lines = 0 for xp_label, xp_data in inputs: for scatter in xp_data.data: args = [] # the data x = scatter.false_positives if self.det: y = scatter.false_negatives else: y = 1.0 - scatter.false_negatives # accumulate counts (useful in log plots) number_of_positives.append(scatter.number_of_positives) number_of_negatives.append(scatter.number_of_negatives) args.append(x * self.xaxis_multiplier) args.append(y * self.yaxis_multiplier) # the label if self.label: #gets the label from user overwritten input label = self.label[n_lines%len(self.label)] else: #make-up label if single_experiment: label = scatter.label else: text = [k for k in (xp_label, scatter.label) if k] label = '-'.join(text) # the line attributes if self.line_attributes: args.append(self.line_attributes[n_lines%len(self.line_attributes)]) else: kwargs['color'] = self.tableau20[n_lines % len(self.tableau20)] kwargs['label'] = label n_lines += 1 ax.plot(*args, **kwargs) if n_lines > 1: ax.legend(loc=self.loc, fontsize=self.legend_fontsize, fancybox=True, framealpha=0.5) # reset log axis if self.xaxis_log: N = min([k for k in number_of_positives if k > 0]) or 100 ax.set_xscale('symlog', linthreshx=1.0/N) if self.yaxis_log: N = min([k for k in number_of_negatives if k > 0]) or 100 ax.set_yscale('symlog', linthreshy=1.0/N) return super(Plotter, self).encode_figure(fig)This is a plotter for ISO/IEC 19795-1:2006(E) standard ROC and DET curves. It is based on the description on the standard, section 10.6 "Graphical presentation of results", pages 35 and 36.