def __init__(
self,
pre_nms_top_n,
post_nms_top_n,
nms_thresh,
min_size,
box_coder=None,
fpn_post_nms_top_n=None,
):
"""
Arguments:
pre_nms_top_n (int)
post_nms_top_n (int)
nms_thresh (float)
min_size (int)
box_coder (BoxCoder)
fpn_post_nms_top_n (int)
"""
super(RPNPostProcessor, self).__init__()
self.pre_nms_top_n = pre_nms_top_n
self.post_nms_top_n = post_nms_top_n
self.nms_thresh = nms_thresh
self.min_size = min_size
if box_coder is None:
box_coder = RBoxCoder(weights=(1.0, 1.0, 1.0, 1.0, 1.0))
self.box_coder = box_coder
if fpn_post_nms_top_n is None:
fpn_post_nms_top_n = post_nms_top_n
self.fpn_post_nms_top_n = fpn_post_nms_top_n
def __init__(self, cfg):
super(RPNModule, self).__init__()
self.cfg = cfg.clone()
anchor_generator = make_anchor_generator(cfg)
in_channels = cfg.MODEL.BACKBONE.OUT_CHANNELS
rpn_head = registry.RPN_HEADS[cfg.MODEL.RPN.RPN_HEAD]
head = rpn_head(cfg, in_channels, 1)
rpn_box_coder = RBoxCoder(weights=(1.0, 1.0, 1.0, 1.0, 1.0))
box_selector_train = make_rpn_postprocessor(cfg,
rpn_box_coder,
is_train=True)
box_selector_test = make_rpn_postprocessor(cfg,
rpn_box_coder,
is_train=False)
loss_evaluator = loss_evaluator_dict[cfg.MODEL.RPN.RPN_HEAD](
cfg, rpn_box_coder)
self.anchor_generator = anchor_generator
self.head = head
self.box_selector_train = box_selector_train
self.box_selector_test = box_selector_test
self.loss_evaluator = loss_evaluator
self.loss_name = loss_name_dict[cfg.MODEL.RPN.RPN_HEAD]
def make_roi_box_post_processor(cfg):
use_fpn = cfg.MODEL.ROI_HEADS.USE_FPN
bbox_reg_weights = cfg.MODEL.ROI_HEADS.RBBOX_REG_WEIGHTS
box_coder = RBoxCoder(weights=bbox_reg_weights)
score_thresh = cfg.MODEL.ROI_HEADS.SCORE_THRESH
nms_thresh = cfg.MODEL.ROI_HEADS.NMS
detections_per_img = cfg.MODEL.ROI_HEADS.DETECTIONS_PER_IMG
postprocessor = PostProcessor(score_thresh, nms_thresh, detections_per_img,
box_coder)
return postprocessor
def __init__(
self, score_thresh=0.05, nms=0.5, detections_per_img=100, box_coder=None
):
"""
Arguments:
score_thresh (float)
nms (float)
detections_per_img (int)
box_coder (BoxCoder)
"""
super(PostProcessor, self).__init__()
self.score_thresh = score_thresh
self.nms = nms
self.detections_per_img = detections_per_img
if box_coder is None:
box_coder = RBoxCoder(weights=(10., 10., 5., 5., 1.))
self.box_coder = box_coder
def make_roi_box_loss_evaluator(cfg):
matcher = Matcher(
cfg.MODEL.ROI_HEADS.FG_IOU_THRESHOLD,
cfg.MODEL.ROI_HEADS.BG_IOU_THRESHOLD,
allow_low_quality_matches=False,
)
bbox_reg_weights = cfg.MODEL.ROI_HEADS.RBBOX_REG_WEIGHTS
box_coder = RBoxCoder(weights=bbox_reg_weights)
fg_bg_sampler = BalancedPositiveNegativeSampler(
cfg.MODEL.ROI_HEADS.BATCH_SIZE_PER_IMAGE,
cfg.MODEL.ROI_HEADS.POSITIVE_FRACTION)
loss_evaluator = FastRCNNLossComputation(matcher, fg_bg_sampler, box_coder)
return loss_evaluator
def __init__(
self, score_thresh=0.02, nms=0.5, detections_per_img=100, box_coder=None, nms_type="remove", shrink_margin=1.4
):
"""
Arguments:
score_thresh (float)
nms (float)
detections_per_img (int)
box_coder (BoxCoder)
nms_type: "remove" or "merge"
"""
super(PostProcessor, self).__init__()
self.score_thresh = score_thresh
self.nms = nms
self.detections_per_img = detections_per_img
if box_coder is None:
box_coder = RBoxCoder(weights=(10., 10., 5., 5., 1.))
self.box_coder = box_coder
self.shrink_margin = shrink_margin
self.nms_fn = boxlist_nms if nms_type == "remove" else cluster_nms
self.update_cls = False
self.update_box = False
def make_roi_box_loss_evaluator(cfg):
matcher = Matcher(
cfg.MODEL.ROI_HEADS.FG_IOU_THRESHOLD,
cfg.MODEL.ROI_HEADS.BG_IOU_THRESHOLD,
allow_low_quality_matches=False,
discard_high_matches=cfg.MODEL.ROI_HEADS.HIGHEST_DISCARD,
highest_threshold=cfg.MODEL.ROI_HEADS.HIGHEST_THRESHOLD,
)
bbox_reg_weights = cfg.MODEL.ROI_HEADS.RBBOX_REG_WEIGHTS
box_coder = RBoxCoder(weights=bbox_reg_weights)
fg_bg_sampler = BalancedPositiveNegativeSampler(
cfg.MODEL.ROI_HEADS.BATCH_SIZE_PER_IMAGE,
cfg.MODEL.ROI_HEADS.POSITIVE_FRACTION)
edge_punished = cfg.MODEL.EDGE_PUNISHED
loss_evaluator = FastRCNNLossComputation(
matcher,
fg_bg_sampler,
box_coder,
edge_punished,
discard_highest=cfg.MODEL.ROI_HEADS.HIGHEST_DISCARD)
return loss_evaluator
def __init__(self,
pre_nms_top_n,
post_nms_top_n,
nms_thresh,
min_size,
nms_type="remove",
box_coder=None,
fpn_post_nms_top_n=None,
base_size=640.,
scale_stack=[0.25, 0.125, 0.0625, 0.03125, 0.015625],
score_thresh=0.1):
"""
Arguments:
pre_nms_top_n (int)
post_nms_top_n (int)
nms_thresh (float)
min_size (int)
box_coder (BoxCoder)
fpn_post_nms_top_n (int)
"""
super(RPNPostProcessor, self).__init__()
self.pre_nms_top_n = pre_nms_top_n
self.post_nms_top_n = post_nms_top_n
self.nms_thresh = nms_thresh
self.min_size = min_size
self.base_size = base_size
self.score_thresh = score_thresh
if box_coder is None:
box_coder = RBoxCoder(weights=(1.0, 1.0, 1.0, 1.0, 1.0))
self.box_coder = box_coder
self.scale_stack = scale_stack
if fpn_post_nms_top_n is None:
fpn_post_nms_top_n = post_nms_top_n
self.fpn_post_nms_top_n = fpn_post_nms_top_n
self.nms_fn = boxlist_nms if nms_type == "remove" else cluster_nms
