cvtools.label_analysis package

Submodules

• cvtools.label_analysis.coco_analysis module
• cvtools.label_analysis.dets_analysis module
• cvtools.label_analysis.size_analysis module

Module contents

class cvtools.label_analysis.COCOAnalysis(img_prefix, ann_file=None)

基类：object

coco-like datasets analysis

cluster_analysis(save_root, name_clusters=('bbox', ), n_clusters=(3, ), by_cat=False)

get_weights_for_balanced_classes(to_file='weighted_samples.pkl')

read_img_or_crop(entry)

stats_objs_per_cat(to_file='objs_per_cat_data.json')

stats_objs_per_img(to_file='stats_num.json')

stats_size_per_cat(to_file='size_per_cat_data.json')

vis_instances(save_root, vis='bbox', vis_cats=None, output_by_cat=False, box_format='x1y1wh')

Visualise bbox and polygon in annotation.

包含某一类的图片上所有类别均会被绘制。

参数:

• save_root (str) -- path for saving image.
• vis (str) -- 'bbox' or 'segmentation'
• vis_cats (list) -- categories to be visualized
• output_by_cat (bool) -- output visual images by category.
• box_format (str) -- 'x1y1wh' or 'polygon'

vis_objs(img, objs, vis='bbox', box_format='x1y1wh')