# boostrapping bootstrapping = 1 # enable bootstrapping extraction or not bootstrapping_save = 1 # save bootstrapped samples directly as a dataset display_bootstrapping = 1 # display bootstrapping extraction # paths gt_path = /xml_path/ # path to xml annotations input_dir = /images_path/ # path of images to bootstrap on output_dir = ${current_dir}# path to save outputs # bootstrapping params gt_pos_matching = .5 # positives have to match at least this much to be considered gt_neg_matching = .4 # negatives can only match positives this much to be negative gt_negneg_matching = .5 # negatives can only match each other this much in a same image gt_min_context = 1.29 # positives must have at least this much context gt_minvisibility = .3 # positives must be at least this much visible gt_min_aspect_ratio = .25 # minimum positives aspect ratio to be considered gt_max_aspect_ratio = .6 # maximum positives aspect ratio to be considered gt_mindims = 2x2 # minimum dimensions of positive boxes to be considered gt_minborders = gt_included = class1,class3# classes to be included for positive extraction gt_mirror_pos = 1 # extract vertically-mirrored samples #gt_scales = .775,.6,.466 gt_neg_max = 20 # maximum number of negatives per classifier and image # negative bootstrapping bootstrapping_max = 1000 # limit to this number of extracted samples (optional) gt_extract_pos = 0 # extract positive samples or not gt_extract_neg = 1 # extract negative samples or not gt_neg_threshold = .01 # minimum confidence of extracted negative samples gt_neg_gt_only = 0 # only extract negatives when positives are present in image input_random = 1 # taking random images is better for bootstrapping gt_name = bootstrap_neg # name of saved bootstrapping dataset bbox_scalings = 1x1 # scaling factors of detected boxes
# positive bootstrapping bootstrapping_max = #10 gt_extract_pos = 1 gt_extract_neg = 0 nthreads = 1 gt_name = bootstrapping_pos