megadetector 5.0.11__py3-none-any.whl → 5.0.13__py3-none-any.whl

megadetector/utils/write_html_image_list.py

@@ -0,0 +1,225 @@
+"""
+
+write_html_image_list.py
+
+Given a list of image file names, writes an HTML file that
+shows all those images, with optional one-line headers above each.
+
+Each "filename" can also be a dict with elements 'filename','title',
+'imageStyle','textStyle', 'linkTarget'
+
+"""
+
+#%% Constants and imports
+
+import os
+import math
+import urllib
+
+from megadetector.utils import path_utils
+
+
+#%% write_html_image_list
+
+def write_html_image_list(filename=None,images=None,options=None):
+    """
+    Given a list of image file names, writes an HTML file that shows all those images, 
+    with optional one-line headers above each.
+
+    Args:
+        filename (str, optional): the .html output file; if None, just returns a valid 
+            options dict
+        images (list, optional): the images to write to the .html file; if None, just returns 
+            a valid options dict.  This can be a flat list of image filenames, or this can
+            be a list of dictionaries with one or more of the following fields:
+                
+            - filename (image filename) (required, all other fields are optional)
+            - imageStyle (css style for this image)
+            - textStyle (css style for the title associated with this image)
+            - title (text label for this image)
+            - linkTarget (URL to which this image should link on click)
+            
+        options (dict, optional): a dict with one or more of the following fields:        
+            
+            - fHtml (file pointer to write to, used for splitting write operations over multiple calls)
+            - headerHtml (html text to include before the image list)
+            - trailerHtml (html text to include after the image list)
+            - defaultImageStyle (default css style for images)
+            - defaultTextStyle (default css style for image titles)
+            - maxFiguresPerHtmlFile (max figures for a single HTML file; overflow will be handled by creating
+              multiple files and a TOC with links)
+            - urlEncodeFilenames (default True, e.g. '#' will be replaced by '%23')
+            - urlEncodeLinkTargets (default True, e.g. '#' will be replaced by '%23')
+            
+    """
+    
+    # returns an options struct
+    if options is None:
+        options = {}
+        
+    if 'fHtml' not in options:
+        options['fHtml'] = -1
+    
+    if 'headerHtml' not in options or options['headerHtml'] is None:
+        options['headerHtml'] = ''        
+    
+    if 'trailerHtml' not in options or options['trailerHtml'] is None:
+        options['trailerHtml'] = ''    
+    
+    if 'defaultTextStyle' not in options or options['defaultTextStyle'] is None:
+        options['defaultTextStyle'] = \
+        "font-family:calibri,verdana,arial;font-weight:bold;font-size:150%;text-align:left;margin:0px;"
+
+    if 'defaultImageStyle' not in options or options['defaultImageStyle'] is None:
+        options['defaultImageStyle'] = \
+        "margin:0px;margin-top:5px;margin-bottom:5px;"
+    
+    if 'urlEncodeFilenames' not in options or options['urlEncodeFilenames'] is None:
+        options['urlEncodeFilenames'] = True
+    
+    if 'urlEncodeLinkTargets' not in options or options['urlEncodeLinkTargets'] is None:
+        options['urlEncodeLinkTargets'] = True
+    
+    # Possibly split the html output for figures into multiple files; Chrome gets sad with
+    # thousands of images in a single tab.        
+    if 'maxFiguresPerHtmlFile' not in options or options['maxFiguresPerHtmlFile'] is None:
+        options['maxFiguresPerHtmlFile'] = math.inf    
+    
+    if filename is None or images is None:
+        return options
+    
+    # images may be a list of images or a list of image/style/title dictionaries, 
+    # enforce that it's the latter to simplify downstream code
+    for iImage,imageInfo in enumerate(images):
+        if isinstance(imageInfo,str):
+            imageInfo = {'filename':imageInfo}
+        if 'filename' not in imageInfo:
+            imageInfo['filename'] = ''
+        if 'imageStyle' not in imageInfo:
+            imageInfo['imageStyle'] = options['defaultImageStyle']
+        if 'title' not in imageInfo:
+            imageInfo['title'] = ''
+        if 'linkTarget' not in imageInfo:
+            imageInfo['linkTarget'] = ''
+        if 'textStyle' not in imageInfo:
+            textStyle = options['defaultTextStyle']
+            imageInfo['textStyle'] = options['defaultTextStyle']
+        images[iImage] = imageInfo            
+    
+    nImages = len(images)
+    
+    # If we need to break this up into multiple files...
+    if nImages > options['maxFiguresPerHtmlFile']:
+    
+        # You can't supply your own file handle in this case
+        if options['fHtml'] != -1:
+            raise ValueError(
+                    'You can''t supply your own file handle if we have to page the image set')
+        
+        figureFileStartingIndices = list(range(0,nImages,options['maxFiguresPerHtmlFile']))
+
+        assert len(figureFileStartingIndices) > 1
+        
+        # Open the meta-output file
+        fMeta = open(filename,'w')
+        
+        # Write header stuff
+        fMeta.write('<html><body>\n')    
+        fMeta.write(options['headerHtml'])        
+        fMeta.write('<table border = 0 cellpadding = 2>\n')
+        
+        for startingIndex in figureFileStartingIndices:
+            
+            iStart = startingIndex
+            iEnd = startingIndex+options['maxFiguresPerHtmlFile']-1;
+            if iEnd >= nImages:
+                iEnd = nImages-1
+            
+            trailer = 'image_{:05d}_{:05d}'.format(iStart,iEnd)
+            localFiguresHtmlFilename = path_utils.insert_before_extension(filename,trailer)
+            fMeta.write('<tr><td>\n')
+            fMeta.write('<p style="padding-bottom:0px;margin-bottom:0px;text-align:left;font-family:''segoe ui'',calibri,arial;font-size:100%;text-decoration:none;font-weight:bold;">')
+            fMeta.write('<a href="{}">Figures for images {} through {}</a></p></td></tr>\n'.format(
+                os.path.basename(localFiguresHtmlFilename),iStart,iEnd))
+            
+            localImages = images[iStart:iEnd+1]
+            
+            localOptions = options.copy();
+            localOptions['headerHtml'] = '';
+            localOptions['trailerHtml'] = '';
+            
+            # Make a recursive call for this image set
+            write_html_image_list(localFiguresHtmlFilename,localImages,localOptions)
+            
+        # ...for each page of images
+        
+        fMeta.write('</table></body>\n')
+        fMeta.write(options['trailerHtml'])
+        fMeta.write('</html>\n')
+        fMeta.close()
+        
+        return options
+        
+    # ...if we have to make multiple sub-pages
+        
+    bCleanupFile = False
+    
+    if options['fHtml'] == -1:
+        bCleanupFile = True;
+        fHtml = open(filename,'w')
+    else:
+        fHtml = options['fHtml']
+        
+    fHtml.write('<html><body>\n')
+    
+    fHtml.write(options['headerHtml'])
+    
+    # Write out images
+    for iImage,image in enumerate(images):
+        
+        title = image['title']
+        imageStyle = image['imageStyle']
+        textStyle = image['textStyle']
+        filename = image['filename']
+        linkTarget = image['linkTarget']
+        
+        # Remove unicode characters
+        title = title.encode('ascii','ignore').decode('ascii')
+        filename = filename.encode('ascii','ignore').decode('ascii')
+        
+        filename = filename.replace('\\','/')
+        if options['urlEncodeFilenames']:            
+            filename = urllib.parse.quote(filename)
+        
+        if len(title) > 0:       
+            fHtml.write(
+                    '<p style="{}">{}</p>\n'\
+                    .format(textStyle,title))            
+
+        linkTarget = linkTarget.replace('\\','/')
+        if options['urlEncodeLinkTargets']:
+            # These are typically absolute paths, so we only want to mess with certain characters
+            linkTarget = urllib.parse.quote(linkTarget,safe=':/')
+            
+        if len(linkTarget) > 0:
+            fHtml.write('<a href="{}">'.format(linkTarget))
+            # imageStyle.append(';border:0px;')
+        
+        fHtml.write('<img src="{}" style="{}">\n'.format(filename,imageStyle))
+        
+        if len(linkTarget) > 0:
+            fHtml.write('</a>')
+            
+        if iImage != len(images)-1:
+            fHtml.write('<br/>')             
+            
+    # ...for each image we need to write
+    
+    fHtml.write(options['trailerHtml'])
+    
+    fHtml.write('</body></html>\n')
+    
+    if bCleanupFile:
+        fHtml.close()    
+
+# ...function

megadetector/visualization/plot_utils.py

@@ -0,0 +1,293 @@
+"""
+
+plot_utils.py
+
+Utility functions for plotting, particularly for plotting confusion matrices
+and precision-recall curves.
+
+"""
+
+#%% Imports
+
+import numpy as np
+
+# This also imports mpl.{cm, axes, colors}
+import matplotlib.figure
+
+
+#%% Plotting functions
+
+def plot_confusion_matrix(matrix,
+                          classes,
+                          normalize=False,
+                          title='Confusion matrix',
+                          cmap=matplotlib.cm.Blues,
+                          vmax=None,
+                          use_colorbar=True,
+                          y_label=True,
+                          fmt= '{:.0f}',
+                          fig=None):
+    """
+    Plots a confusion matrix.
+
+    Args:
+        matrix (np.ndarray): shape [num_classes, num_classes], confusion matrix
+            where rows are ground-truth classes and columns are predicted classes
+        classes (list of str): class names for each row/column
+        normalize (bool, optional): whether to perform row-wise normalization; 
+            by default, assumes values in the confusion matrix are percentages
+        title (str, optional): figure title
+        cmap (matplotlib.colors.colormap): colormap for cell backgrounds
+        vmax (float, optional), value corresponding to the largest value of the colormap;
+            if None, the maximum value in [matrix] will be used
+        use_colorbar (bool, optional): whether to show colorbar
+        y_label (bool, optional): whether to show class names on the y axis
+        fmt (str): format string for rendering numeric values
+        fig (Figure): existing figure to which we should render, otherwise creates
+            a new figure
+        
+    Returns: 
+        matplotlib.figure.Figure: the figure we rendered to or created
+    """
+    
+    num_classes = matrix.shape[0]
+    assert matrix.shape[1] == num_classes
+    assert len(classes) == num_classes
+
+    normalized_matrix = matrix.astype(np.float64) / (
+        matrix.sum(axis=1, keepdims=True) + 1e-7)
+    if normalize:
+        matrix = normalized_matrix
+
+    fig_h = 3 + 0.3 * num_classes
+    fig_w = fig_h
+    if use_colorbar:
+        fig_w += 0.5
+
+    if fig is None:
+        fig = matplotlib.figure.Figure(figsize=(fig_w, fig_h), tight_layout=True)
+    ax = fig.subplots(1, 1)
+    im = ax.imshow(normalized_matrix, interpolation='nearest', cmap=cmap, vmax=vmax)
+    ax.set_title(title)
+
+    if use_colorbar:
+        cbar = fig.colorbar(im, fraction=0.046, pad=0.04,
+                            ticks=[0.0, 0.25, 0.5, 0.75, 1.0])
+        cbar.set_ticklabels(['0%', '25%', '50%', '75%', '100%'])
+
+    tick_marks = np.arange(num_classes)
+    ax.set_xticks(tick_marks)
+    ax.set_yticks(tick_marks)
+    ax.set_xticklabels(classes, rotation=90)
+    ax.set_xlabel('Predicted class')
+
+    if y_label:
+        ax.set_yticklabels(classes)
+        ax.set_ylabel('Ground-truth class')
+
+    for i, j in np.ndindex(matrix.shape):
+        v = matrix[i, j]        
+        ax.text(j, i, fmt.format(v),
+                horizontalalignment='center',
+                verticalalignment='center',
+                color='white' if normalized_matrix[i, j] > 0.5 else 'black')
+
+    return fig
+
+# ...def plot_confusion_matrix(...)
+
+
+def plot_precision_recall_curve(precisions, 
+                                recalls, 
+                                title='Precision/recall curve',
+                                xlim=(0.0,1.05),
+                                ylim=(0.0,1.05)):
+    """
+    Plots a precision/recall curve given lists of (ordered) precision
+    and recall values.
+
+    Args:
+        precisions (list of float): precision for corresponding recall values,
+            should have same length as [recalls].
+        recalls (list of float): recall for corresponding precision values,
+            should have same length as [precisions].
+        title (str, optional): plot title
+        xlim (tuple, optional): x-axis limits as a length-2 tuple
+        ylim (tuple, optional): y-axis limits as a length-2 tuple
+
+    Returns: 
+        matplotlib.figure.Figure: the (new) figure
+    """
+    
+    assert len(precisions) == len(recalls)
+
+    fig = matplotlib.figure.Figure(tight_layout=True)
+    ax = fig.subplots(1, 1)
+    ax.step(recalls, precisions, color='b', alpha=0.2, where='post')
+    ax.fill_between(recalls, precisions, alpha=0.2, color='b', step='post')
+
+    try:
+        ax.set(x_label='Recall', y_label='Precision', title=title)
+        ax.set(x_lim=xlim, y_lim=ylim)    
+    # 
+    except Exception:
+        ax.set_xlabel('Recall')
+        ax.set_ylabel('Precision')
+        ax.set_title(title)
+        ax.set_xlim(xlim[0],xlim[1])
+        ax.set_ylim(ylim[0],ylim[1])
+        
+    return fig
+
+
+def plot_stacked_bar_chart(data, series_labels=None, col_labels=None,
+                           x_label=None, y_label=None, log_scale=False):
+    """
+    Plot a stacked bar chart, for plotting e.g. species distribution across locations.
+    
+    Reference: https://stackoverflow.com/q/44309507
+
+    Args:
+        data (np.ndarray or list of list): data to plot; rows (series) are species, columns
+            are locations
+        series_labels (list of str, optional): series labels, typically species names
+        col_labels (list of str, optional): column labels, typically location names
+        x_label (str, optional): x-axis label
+        y_label (str, optional): y-axis label
+        log_scale (bool, optional) whether to plot the y axis in log-scale
+
+    Returns: 
+        matplotlib.figure.Figure: the (new) figure
+    """
+    
+    data = np.asarray(data)
+    num_series, num_columns = data.shape
+    ind = np.arange(num_columns)
+
+    fig = matplotlib.figure.Figure(tight_layout=True)
+    ax = fig.subplots(1, 1)
+    colors = matplotlib.cm.rainbow(np.linspace(0, 1, num_series))
+
+    # stacked bar charts are made with each segment starting from a y position
+    cumulative_size = np.zeros(num_columns)
+    for i, row_data in enumerate(data):
+        ax.bar(ind, row_data, bottom=cumulative_size, label=series_labels[i],
+               color=colors[i])
+        cumulative_size += row_data
+
+    if col_labels and len(col_labels) < 25:
+        ax.set_xticks(ind)
+        ax.set_xticklabels(col_labels, rotation=90)
+    elif col_labels:
+        ax.set_xticks(list(range(0, len(col_labels), 20)))
+        ax.set_xticklabels(col_labels, rotation=90)
+
+    if x_label is not None:
+        ax.set_xlabel(x_label)
+    if y_label is not None:
+        ax.set_ylabel(y_label)
+    if log_scale:
+        ax.set_yscale('log')
+
+    # To fit the legend in, shrink current axis by 20%
+    box = ax.get_position()
+    ax.set_position([box.x0, box.y0, box.width * 0.8, box.height])
+
+    # Put a legend to the right of the current axis
+    ax.legend(loc='center left', bbox_to_anchor=(0.99, 0.5), frameon=False)
+
+    return fig
+
+
+def calibration_ece(true_scores, pred_scores, num_bins):
+    r"""
+    Expected calibration error (ECE) as defined in equation (3) of
+    Guo et al. "On Calibration of Modern Neural Networks." (2017).
+
+    Implementation modified from sklearn.calibration.calibration_curve()
+    in order to implement ECE calculation. See:
+    
+    https://github.com/scikit-learn/scikit-learn/issues/18268
+
+    Args:
+        true_scores (list of int): true values, length N, binary-valued (0 = neg, 1 = pos)
+        pred_scores (list of float): predicted confidence values, length N, pred_scores[i] is the 
+            predicted confidence that example i is positive
+        num_bins (int): number of bins to use (`M` in eq. (3) of Guo 2017)
+
+    Returns:
+        tuple: a length-three tuple containing:
+            - accs: np.ndarray, shape [M], type float64, accuracy in each bin,
+              M <= num_bins because bins with no samples are not returned        
+            - confs: np.ndarray, shape [M], type float64, mean model confidence in
+              each bin
+            - ece: float, expected calibration error
+    """
+    
+    assert len(true_scores) == len(pred_scores)
+
+    bins = np.linspace(0., 1. + 1e-8, num=num_bins + 1)
+    binids = np.digitize(pred_scores, bins) - 1
+
+    bin_sums = np.bincount(binids, weights=pred_scores, minlength=len(bins))
+    bin_true = np.bincount(binids, weights=true_scores, minlength=len(bins))
+    bin_total = np.bincount(binids, minlength=len(bins))
+
+    nonzero = bin_total != 0
+    accs = bin_true[nonzero] / bin_total[nonzero]
+    confs = bin_sums[nonzero] / bin_total[nonzero]
+
+    weights = bin_total[nonzero] / len(true_scores)
+    ece = np.abs(accs - confs) @ weights
+    return accs, confs, ece
+
+
+def plot_calibration_curve(true_scores, pred_scores, num_bins,
+                           name='calibration', plot_perf=True, plot_hist=True,
+                           ax=None, **fig_kwargs):
+    """
+    Plots a calibration curve.
+
+    Args:
+        true_scores (list of int): true values, length N, binary-valued (0 = neg, 1 = pos)
+        pred_scores (list of float): predicted confidence values, length N, pred_scores[i] is the 
+            predicted confidence that example i is positive
+        num_bins (int): number of bins to use (`M` in eq. (3) of Guo 2017)
+        name (str, optional): label in legend for the calibration curve
+        plot_perf (bool, optional): whether to plot y=x line indicating perfect calibration
+        plot_hist (bool, optional): whether to plot histogram of counts
+        ax (Axes, optional): if given then no legend is drawn, and fig_kwargs are ignored
+        fig_kwargs (dict, optional): only used if [ax] is None
+
+    Returns:
+        matplotlib.figure.Figure: the (new) figure
+    """
+    
+    accs, confs, ece = calibration_ece(true_scores, pred_scores, num_bins)
+
+    created_fig = False
+    if ax is None:
+        created_fig = True
+        fig = matplotlib.figure.Figure(**fig_kwargs)
+        ax = fig.subplots(1, 1)
+    ax.plot(confs, accs, 's-', label=name)  # 's-': squares on line
+    ax.set(xlabel='Model confidence', ylabel='Actual accuracy',
+           title=f'Calibration plot (ECE: {ece:.02g})')
+    ax.set(xlim=[-0.05, 1.05], ylim=[-0.05, 1.05])
+    if plot_perf:
+        ax.plot([0, 1], [0, 1], color='black', label='perfect calibration')
+    ax.grid(True)
+
+    if plot_hist:
+        ax1 = ax.twinx()
+        bins = np.linspace(0., 1. + 1e-8, num=num_bins + 1)
+        counts = ax1.hist(pred_scores, alpha=0.5, label='histogram of examples',
+                          bins=bins, color='tab:red')[0]
+        max_count = np.max(counts)
+        ax1.set_ylim([-0.05 * max_count, 1.05 * max_count])
+        ax1.set_ylabel('Count')
+
+    if created_fig:
+        fig.legend(loc='upper left', bbox_to_anchor=(0.15, 0.85))
+
+    return ax.figure