# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved. from core.utils.registry import Registry BACKBONES = Registry() RPN_HEADS = Registry() ROI_BOX_FEATURE_EXTRACTORS = Registry() ROI_BOX_PREDICTOR = Registry() ROI_KEYPOINT_FEATURE_EXTRACTORS = Registry() ROI_KEYPOINT_PREDICTOR = Registry() ROI_MASK_FEATURE_EXTRACTORS = Registry() ROI_MASK_PREDICTOR = Registry()
from copy import deepcopy
from core.utils.registry import Registry
from pycocotools.coco import COCO
BOXES_FILTERS = Registry()
@BOXES_FILTERS.register('all')
def filter_all(box, im_w, im_h):
return True
@BOXES_FILTERS.register('center')
def filter_center(box, im_w, im_h):
x, y, w, h = box
cx, cy = x + w/2.0, y + h/2.0
return im_w*0.25 <= cx <= im_w*0.75 and im_h*0.25 <= cy <= im_h*0.75
@BOXES_FILTERS.register('border')
def filter_border(box, im_w, im_h):
x, y, w, h = box
cx, cy = x + w / 2.0, y + h / 2.0
return cx <= im_w*0.25 or cx >= im_w*0.75 or cy <= im_h*0.25 or cy >= im_h*0.75
@BOXES_FILTERS.register('left')
def filter_left(box, im_w, im_h):
x, y, w, h = box
cx, cy = x + w / 2.0, y + h / 2.0
return cx <= im_w*0.25
import itertools
import torch
import torch.nn.functional as F
from torch import nn
from core.layers import ConvBlock, NoopLayer
from core.utils.registry import Registry
from core.modeling.rpn.utils import meshgrid
MONTAGE_BOXES = Registry()
# register levels
MONTAGE_LEVELS = {
'type-1': [0, 1, 2, 3, 4],
'type-2': [0, 1, 2, 3, 4, 2, 3, 4],
'type-3': [0, 1, 2, 3, 4],
'type-4': [0, 1, 2, 3, 4, 2, 3, 4],
'type-5': [0, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 4],
'type-6': [0, 1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 4],
'type-7': [0, 0, 0, 1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 4],
'type-8': [0, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 4, 4, 4],
'type-9': [0, 1, 2, 2, 3, 4, 3, 4, 3, 4, 3, 4, 3, 4],
'type-stair1': [1, 1, 1, 1],
'type-stair2': [2, 2, 2, 2],
'type-stair3': [3, 3, 3, 3],
'type-stair4': [4, 4, 4, 4],
'type-sbs1': [1, 1, 1, 1],
'type-sbs2': [2, 2, 2, 2],
'type-sbs3': [3, 3, 3, 3],
'type-sbs4': [4, 4, 4, 4],
'type-grid1': [1, 1, 1, 1],
'type-grid2': [2, 2, 2, 2],
import torch
import torch.nn.functional as F
from torch import nn
from core.layers import ConvBlock, CollectionConvBlock
from core.utils.registry import Registry
# ATTENTION: all FE modules should provide attribute: `out_channels`
FE_MODULES = Registry()
class FEModule(torch.nn.Module):
"""
Feature embedding module
"""
def __init__(self, cfg, in_channels):
super(FEModule, self).__init__()
num_convs = cfg.MODEL.RETINAPACK.FE.NUM_CONVS
self.out_channels = cfg.MODEL.RETINAPACK.FE.OUT_CHANNELS
self.reduction = ConvBlock(in_channels,
self.out_channels,
kernel_size=1,
stride=1,
padding=0)
self.conv_blocks = nn.ModuleList([
ConvBlock(self.out_channels,
self.out_channels,
kernel_size=3,
use_dcn=False,
use_gn=True,
from core.backbone import ResNet
from core.utils.registry import Registry
BACKBONES = Registry()
@BACKBONES.register("ResNet50")
@BACKBONES.register("ResNet101")
def build_resnet_backbone(cfg):
return ResNet(*cfg.backbone.args)
@BACKBONES.register("DarkNet53")
def build_darknet_backbone(cfg):
print("DarkNet53 hasn't been implemented yet, using ResNet instead")
return ResNet(*cfg.backbone.args)
def build_backbone(cfg):
return BACKBONES[cfg.backbone.name](cfg)
out_channels=out_channels,
num_groups=num_groups,
stride_in_1x1=stride_in_1x1,
stride=stride,
dilation=dilation,
norm_func=group_norm,
dcn_config=dcn_config)
class StemWithGN(BaseStem):
def __init__(self, cfg):
super(StemWithGN, self).__init__(cfg, norm_func=group_norm)
_TRANSFORMATION_MODULES = Registry({
"BottleneckWithFixedBatchNorm": BottleneckWithFixedBatchNorm,
"BottleneckWithGN": BottleneckWithGN,
})
_STEM_MODULES = Registry({
"StemWithFixedBatchNorm": StemWithFixedBatchNorm,
"StemWithGN": StemWithGN,
})
_STAGE_SPECS = Registry({
"R-50-C4": ResNet50StagesTo4,
"R-50-C5": ResNet50StagesTo5,
"R-101-C4": ResNet101StagesTo4,
"R-101-C5": ResNet101StagesTo5,
"R-50-FPN": ResNet50FPNStagesTo5,
"R-50-FPN-RETINANET": ResNet50FPNStagesTo5,
"R-101-FPN": ResNet101FPNStagesTo5,
# encoding: utf-8 """ @author: liaoxingyu @contact: [email protected] """ import torch from core.utils.registry import Registry META_ARCH_REGISTRY = Registry("META_ARCH") # noqa F401 isort:skip META_ARCH_REGISTRY.__doc__ = """ Registry for meta-architectures, i.e. the whole model. The registered object will be called with `obj(cfg)` and expected to return a `nn.Module` object. """ def build_model(cfg): """ Build the whole model architecture, defined by ``cfg.MODEL.META_ARCHITECTURE``. Note that it does not load any weights from ``cfg``. """ meta_arch = cfg.MODEL.META_ARCHITECTURE model = META_ARCH_REGISTRY.get(meta_arch)(cfg) model.to(torch.device(cfg.MODEL.DEVICE)) return model
* Single-Path NAS Pixel1
- Single-Path NAS: Designing Hardware-Efficient ConvNets - https://arxiv.org/abs/1904.02877
* And likely more...
Hacked together by Ross Wightman
"""
from .efficientnets.efficientnet_builder import *
from .efficientnets.conv2d_layers import select_conv2d
from .efficientnets.constants import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD, \
IMAGENET_INCEPTION_MEAN, IMAGENET_INCEPTION_STD
from core.utils.registry import Registry
from core.layers import FrozenBatchNorm2d
EFN = Registry()
_DEBUG = False
def _cfg(url='', **kwargs):
return {
'url': url,
'num_classes': 1000,
'input_size': (3, 224, 224),
'pool_size': (7, 7),
'crop_pct': 0.875,
'interpolation': 'bicubic',
'mean': IMAGENET_DEFAULT_MEAN,
'std': IMAGENET_DEFAULT_STD,
'first_conv': 'conv_stem',
'classifier': 'classifier',
)
logger.info("pattern: {}, old_key: {}, new_key: {}".format(
pattern, old_key, new_key
))
state_dict[new_key] = state_dict[old_key]
del state_dict[old_key]
return state_dict
_C2_STAGE_NAMES = {
"R-50": ["1.2", "2.3", "3.5", "4.2"],
"R-101": ["1.2", "2.3", "3.22", "4.2"],
"R-152": ["1.2", "2.7", "3.35", "4.2"],
}
C2_FORMAT_LOADER = Registry()
@C2_FORMAT_LOADER.register("R-50-C4")
@C2_FORMAT_LOADER.register("R-50-C5")
@C2_FORMAT_LOADER.register("R-101-C4")
@C2_FORMAT_LOADER.register("R-101-C5")
@C2_FORMAT_LOADER.register("R-50-FPN")
@C2_FORMAT_LOADER.register("R-50-FPN-RETINANET")
@C2_FORMAT_LOADER.register("R-101-FPN")
@C2_FORMAT_LOADER.register("R-101-FPN-RETINANET")
@C2_FORMAT_LOADER.register("R-152-FPN")
def load_resnet_c2_format(cfg, f):
state_dict = _load_c2_pickled_weights(f)
conv_body = cfg.MODEL.BACKBONE.CONV_BODY
arch = conv_body.replace("-C4", "").replace("-C5", "").replace("-SFPN", "").replace("-FPN", "")