def __init__(
self,
backbone,
n_channel_backbone=5,
num_classes=None,
# transform parameters
min_size=800,
max_size=1333,
#min_size=720, max_size=1280,
image_mean=None,
image_std=None,
# RPN parameters
rpn_anchor_generator=None,
rpn_head=None,
rpn_pre_nms_top_n_train=2000,
rpn_pre_nms_top_n_test=1000,
rpn_post_nms_top_n_train=2000,
rpn_post_nms_top_n_test=1000,
rpn_nms_thresh=0.5,
rpn_fg_iou_thresh=0.7,
rpn_bg_iou_thresh=0.3,
rpn_batch_size_per_image=256,
rpn_positive_fraction=0.5,
anchor_sizes=[32, 64, 128, 256, 512],
# Box parameters
box_roi_pool=None,
box_head=None,
box_predictor=None,
box_score_thresh=0.05,
box_nms_thresh=0.4,
box_detections_per_img=30,
box_fg_iou_thresh=0.5,
box_bg_iou_thresh=0.5,
box_batch_size_per_image=512,
box_positive_fraction=0.25,
bbox_reg_weights=None,
weight_loss=False,
use_soft_nms=False,
use_context=False,
use_track_branch=False):
if not hasattr(backbone, "out_channels"):
raise ValueError(
"backbone should contain an attribute out_channels "
"specifying the number of output channels (assumed to be the "
"same for all the levels)")
assert isinstance(rpn_anchor_generator, (AnchorGenerator, type(None)))
assert isinstance(box_roi_pool, (MultiScaleRoIAlign, type(None)))
if num_classes is not None:
if box_predictor is not None:
raise ValueError(
"num_classes should be None when box_predictor is specified"
)
else:
if box_predictor is None:
raise ValueError(
"num_classes should not be None when box_predictor "
"is not specified")
out_channels = backbone.out_channels
if rpn_anchor_generator is None:
ratios = ((0.5, 1.0, 2.0), )
aspect_ratios = ratios * len(anchor_sizes)
rpn_anchor_generator = AnchorGenerator(anchor_sizes, aspect_ratios)
if rpn_head is None:
rpn_head = RPNHead(
out_channels,
rpn_anchor_generator.num_anchors_per_location()[0])
rpn_pre_nms_top_n = dict(training=rpn_pre_nms_top_n_train,
testing=rpn_pre_nms_top_n_test)
rpn_post_nms_top_n = dict(training=rpn_post_nms_top_n_train,
testing=rpn_post_nms_top_n_test)
rpn = RegionProposalNetwork(rpn_anchor_generator, rpn_head,
rpn_fg_iou_thresh, rpn_bg_iou_thresh,
rpn_batch_size_per_image,
rpn_positive_fraction, rpn_pre_nms_top_n,
rpn_post_nms_top_n, rpn_nms_thresh,
weight_loss)
if box_roi_pool is None:
box_roi_pool = MultiScaleRoIAlign(featmap_names=[0, 1, 2, 3],
output_size=7,
sampling_ratio=2)
if n_channel_backbone == 6:
box_roi_pool = MultiScaleRoIAlign(
featmap_names=[0, 1, 2, 3, 4],
output_size=7,
sampling_ratio=2)
representation_size1 = 1024
representation_size2 = 1024
track_embedding_size = 1024
if box_head is None:
resolution = box_roi_pool.output_size[0]
if use_context:
box_head = TwoMLPHead(2 * out_channels * resolution**2,
representation_size1,
representation_size2)
else:
box_head = TwoMLPHead(out_channels * resolution**2,
representation_size1,
representation_size2)
if use_track_branch:
if use_context:
track_embedding = TwoMLPHead(2 * out_channels * resolution**2,
representation_size1,
track_embedding_size)
else:
track_embedding = TwoMLPHead(out_channels * resolution**2,
representation_size1,
track_embedding_size)
else:
track_embedding = None
if box_predictor is None:
box_predictor = FastRCNNPredictor(representation_size1,
num_classes)
if num_classes > 2:
use_soft_nms = False
roi_heads = RoIHeads(
# Box
box_roi_pool,
box_head,
box_predictor,
box_fg_iou_thresh,
box_bg_iou_thresh,
box_batch_size_per_image,
box_positive_fraction,
bbox_reg_weights,
box_score_thresh,
box_nms_thresh,
box_detections_per_img,
weight_loss=weight_loss,
use_soft_nms=use_soft_nms,
use_context=use_context)
if use_track_branch:
track_heads = TrackHeads(box_roi_pool,
box_head,
box_predictor,
box_fg_iou_thresh,
box_bg_iou_thresh,
box_batch_size_per_image,
box_positive_fraction,
bbox_reg_weights,
weight_loss=False,
use_context=False,
track_embedding=track_embedding)
else:
track_heads = None
if image_mean is None:
image_mean = [0.485, 0.456, 0.406]
if image_std is None:
image_std = [0.229, 0.224, 0.225]
transform = GeneralizedRCNNTransform(min_size, max_size, image_mean,
image_std)
super(FasterRCNN, self).__init__(backbone, rpn, roi_heads, track_heads,
transform, n_channel_backbone)
def __init__(self, backbone, num_classes=None,
# transform parameters
min_size=800, max_size=1333,
image_mean=None, image_std=None,
# RPN parameters
rpn_anchor_generator=None, rpn_head=None,
rpn_pre_nms_top_n_train=2000, rpn_pre_nms_top_n_test=1000,
rpn_post_nms_top_n_train=2000, rpn_post_nms_top_n_test=1000,
rpn_nms_thresh=0.7,
rpn_fg_iou_thresh=0.7, rpn_bg_iou_thresh=0.3,
rpn_batch_size_per_image=256, rpn_positive_fraction=0.5,
# Box parameters
box_roi_pool=None, box_head=None, box_predictor=None,
box_score_thresh=0.05, box_nms_thresh=0.5, box_detections_per_img=100,
box_fg_iou_thresh=0.5, box_bg_iou_thresh=0.5,
box_batch_size_per_image=512, box_positive_fraction=0.25,
bbox_reg_weights=None):
if not hasattr(backbone, "out_channels"):
raise ValueError(
"backbone should contain an attribute out_channels "
"specifying the number of output channels (assumed to be the "
"same for all the levels)")
assert isinstance(rpn_anchor_generator, (AnchorGenerator, type(None)))
assert isinstance(box_roi_pool, (MultiScaleRoIAlign, type(None)))
if num_classes is not None:
if box_predictor is not None:
raise ValueError("num_classes should be None when box_predictor is specified")
else:
if box_predictor is None:
raise ValueError("num_classes should not be None when box_predictor "
"is not specified")
out_channels = backbone.out_channels
if rpn_anchor_generator is None:
anchor_sizes = ((32,), (64,), (128,), (256,), (512,))
aspect_ratios = ((0.5, 1.0, 2.0),) * len(anchor_sizes)
rpn_anchor_generator = AnchorGenerator(
anchor_sizes, aspect_ratios
)
if rpn_head is None:
rpn_head = RPNHead(
out_channels, rpn_anchor_generator.num_anchors_per_location()[0]
)
rpn_pre_nms_top_n = dict(training=rpn_pre_nms_top_n_train, testing=rpn_pre_nms_top_n_test)
rpn_post_nms_top_n = dict(training=rpn_post_nms_top_n_train, testing=rpn_post_nms_top_n_test)
rpn = RegionProposalNetwork(
rpn_anchor_generator, rpn_head,
rpn_fg_iou_thresh, rpn_bg_iou_thresh,
rpn_batch_size_per_image, rpn_positive_fraction,
rpn_pre_nms_top_n, rpn_post_nms_top_n, rpn_nms_thresh)
if box_roi_pool is None:
box_roi_pool = MultiScaleRoIAlign(
featmap_names=['0', '1', '2', '3'],
output_size=7,
sampling_ratio=2)
if box_head is None:
resolution = box_roi_pool.output_size[0]
representation_size = 1024
box_head = TwoMLPHead(
out_channels * resolution ** 2,
representation_size)
if box_predictor is None:
representation_size = 1024
box_predictor = FastRCNNPredictor(
representation_size,
num_classes)
roi_heads = RoIHeads(
# Box
box_roi_pool, box_head, box_predictor,
box_fg_iou_thresh, box_bg_iou_thresh,
box_batch_size_per_image, box_positive_fraction,
bbox_reg_weights,
box_score_thresh, box_nms_thresh, box_detections_per_img)
if image_mean is None:
image_mean = [0.485, 0.456, 0.406]
if image_std is None:
image_std = [0.229, 0.224, 0.225]
transform = GeneralizedRCNNTransform(min_size, max_size, image_mean, image_std)
super(FasterRCNN, self).__init__(backbone, rpn, roi_heads, transform)
def __init__(
self,
backbone,
# transform parameters
min_size=800,
max_size=2000,
image_mean=None,
image_std=None,
# RPN parameters
rpn_anchor_generator=None,
rpn_head=None,
rpn_pre_nms_top_n_train=2000,
rpn_pre_nms_top_n_test=1000,
rpn_post_nms_top_n_train=2000,
rpn_post_nms_top_n_test=1000,
rpn_nms_thresh=0.7,
rpn_fg_iou_thresh=0.7,
rpn_bg_iou_thresh=0.3,
rpn_batch_size_per_image=256,
rpn_positive_fraction=0.5,
weight_loss=False,
# Additional NMS parameters
last_nms_iou_thres=0.3,
n_max_det=30):
super(RPN, self).__init__()
if not hasattr(backbone, "out_channels"):
raise ValueError(
"backbone should contain an attribute out_channels "
"specifying the number of output channels (assumed to be the "
"same for all the levels)")
assert isinstance(rpn_anchor_generator, (AnchorGenerator, type(None)))
self.backbone = backbone
out_channels = backbone.out_channels
if rpn_anchor_generator is None:
anchor_sizes = ((32, ), (64, ), (128, ), (256, ), (512, ))
aspect_ratios = ((0.5, 1.0, 2.0), ) * len(anchor_sizes)
rpn_anchor_generator = AnchorGenerator(anchor_sizes, aspect_ratios)
if rpn_head is None:
rpn_head = RPNHead(
out_channels,
rpn_anchor_generator.num_anchors_per_location()[0])
rpn_pre_nms_top_n = dict(training=rpn_pre_nms_top_n_train,
testing=rpn_pre_nms_top_n_test)
rpn_post_nms_top_n = dict(training=rpn_post_nms_top_n_train,
testing=rpn_post_nms_top_n_test)
rpn = RegionProposalNetwork(rpn_anchor_generator,
rpn_head,
rpn_fg_iou_thresh,
rpn_bg_iou_thresh,
rpn_batch_size_per_image,
rpn_positive_fraction,
rpn_pre_nms_top_n,
rpn_post_nms_top_n,
rpn_nms_thresh,
weight_loss=weight_loss)
self.rpn = rpn
self.last_nms_iou_thres = last_nms_iou_thres
self.n_max_det = n_max_det
if image_mean is None:
image_mean = [0.485, 0.456, 0.406]
if image_std is None:
image_std = [0.229, 0.224, 0.225]
transform = GeneralizedRCNNTransform(min_size, max_size, image_mean,
image_std)
self.transform = transform