def get_coco_dataset():
"""A dummy COCO dataset that includes only the 'classes' field."""
ds = AttrDict()
# classes = [
# '__background__', 'person', 'bicycle', 'car', 'motorcycle', 'airplane',
# 'bus', 'train', 'truck', 'boat', 'traffic light', 'fire hydrant',
# 'stop sign', 'parking meter', 'bench', 'bird', 'cat', 'dog', 'horse',
# 'sheep', 'cow', 'elephant', 'bear', 'zebra', 'giraffe', 'backpack',
# 'umbrella', 'handbag', 'tie', 'suitcase', 'frisbee', 'skis',
# 'snowboard', 'sports ball', 'kite', 'baseball bat', 'baseball glove',
# 'skateboard', 'surfboard', 'tennis racket', 'bottle', 'wine glass',
# 'cup', 'fork', 'knife', 'spoon', 'bowl', 'banana', 'apple', 'sandwich',
# 'orange', 'broccoli', 'carrot', 'hot dog', 'pizza', 'donut', 'cake',
# 'chair', 'couch', 'potted plant', 'bed', 'dining table', 'toilet', 'tv',
# 'laptop', 'mouse', 'remote', 'keyboard', 'cell phone', 'microwave',
# 'oven', 'toaster', 'sink', 'refrigerator', 'book', 'clock', 'vase',
# 'scissors', 'teddy bear', 'hair drier', 'toothbrush'
# ]
classes = [
'__background__', '11', '12', '4', '10', '15', '14',
'8', '3', '1', '13', '9', '7',
'17', '6', '18', '19', '16', '20', '2',
'5'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def _decode_cfg_value(v):
"""Decodes a raw config value (e.g., from a yaml config files or command
line argument) into a Python object.
"""
# Configs parsed from raw yaml will contain dictionary keys that need to be
# converted to AttrDict objects
if isinstance(v, dict):
return AttrDict(v)
# All remaining processing is only applied to strings
if not isinstance(v, six.string_types):
return v
# Try to interpret `v` as a:
# string, number, tuple, list, dict, boolean, or None
try:
v = literal_eval(v)
# The following two excepts allow v to pass through when it represents a
# string.
#
# Longer explanation:
# The type of v is always a string (before calling literal_eval), but
# sometimes it *represents* a string and other times a data structure, like
# a list. In the case that v represents a string, what we got back from the
# yaml parser is 'foo' *without quotes* (so, not '"foo"'). literal_eval is
# ok with '"foo"', but will raise a ValueError if given 'foo'. In other
# cases, like paths (v = 'foo/bar' and not v = '"foo/bar"'), literal_eval
# will raise a SyntaxError.
except ValueError:
pass
except SyntaxError:
pass
return v
def merge_dicts(dict_a, dict_b):
from ast import literal_eval
for key, value in dict_a.items():
if key not in dict_b:
raise KeyError('Invalid key in config file: {}'.format(key))
if type(value) is dict:
dict_a[key] = value = AttrDict(value)
if isinstance(value, str):
try:
value = literal_eval(value)
except BaseException:
pass
# The types must match, too.
old_type = type(dict_b[key])
if old_type is not type(value) and value is not None:
raise ValueError(
'Type mismatch ({} vs. {}) for config key: {}'.format(
type(dict_b[key]), type(value), key))
# Recursively merge dicts.
if isinstance(value, AttrDict):
try:
merge_dicts(dict_a[key], dict_b[key])
except BaseException:
raise Exception('Error under config key: {}'.format(key))
else:
dict_b[key] = value
def get_steal_oil_class14_dataset():
"""A dummy COCO dataset that includes only the 'classes' field."""
ds = AttrDict()
classes = [
'__background__', 'suv', 'forklift', 'digger', 'car', 'bus','tanker', 'person','minitruck','minibus','truckbig','trucksmall','tricycle','bicycle'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_apollo_dataset():
ds = AttrDict()
classes = [
'__background__', 'ignore', 'pedestrian', 'motorcyclist', 'car', 'bus',
'truck', 'tricyclelist', 'van', 'cyclist', 'trafficcone', 'barrowlist'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_common_dataset():
"""A dummy COCO dataset that includes only the 'classes' field."""
ds = AttrDict()
classes = [
'__background__', 'person', 'animal', 'rider', 'motorcycle', 'bicycle', 'autorickshaw', 'car', 'truck', 'bus', 'caravan', 'trailer', 'train', 'vehicle fallback'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_kitti_dataset():
ds = AttrDict()
classes = [
'__background__', 'person', 'rider', 'car', 'truck', 'bus', 'carvanan',
'trailer', 'train', 'motorcycle', 'bicycle'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_kitti_dataset():
"""A dummy kitti dataset that includes only the 'classes' field."""
ds = AttrDict()
classes = [
'__background__', 'cyclist', 'pedestrian', 'car', 'tram', 'truck', 'van', 'misc'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_drive_dataset():
"""A dummy COCO dataset that includes only the 'classes' field."""
ds = AttrDict()
classes = [
'__background__', 'car', 'pedestrian', 'mover', 'traffic light', 'shaft', 'traffic sign'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_ade_dataset():
"""A dummy COCO dataset that includes only the 'classes' field."""
ds = AttrDict()
PATH = os.path.dirname(__file__)
classes = get_classes(os.path.join(PATH, "instanceInfo100_train.txt"))
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_rsna_dataset():
"""A dummy COCO dataset that includes only the 'classes' field."""
ds = AttrDict()
classes = [
'__background__', 'opticapy'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_objectness_dataset():
"""A dummy objectness dataset that includes only the 'classes' field.
This dataset has only two categories: background and object."""
ds = AttrDict()
classes = ['__background__', 'object']
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_illbuild_class11_dataset():
"""A dummy COCO dataset that includes only the 'classes' field."""
ds = AttrDict()
classes = [
'__background__', 'autotruck', 'crane', 'forklift', 'mixerTruck',
'person', 'colorPlate', 'pit', 'bricksPile', 'mound', 'car'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_steal_oil_class8_dataset():
"""A dummy COCO dataset that includes only the 'classes' field."""
ds = AttrDict()
classes = [
'__background__', 'autotruck', 'forklift', 'digger', 'car', 'bus',
'tanker', 'person'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def cfg_from_file(filename):
"""Load a config file and merge it into the default options."""
import yaml
with open(filename, 'r') as f:
yaml_cfg = yaml.load(f)
yaml_cfg = _config_mapping_rules(yaml_cfg)
yaml_cfg = AttrDict(yaml_cfg)
_merge_a_into_b(yaml_cfg, __C)
def get_miotcd_dataset():
"""A dummy MIOTCD dataset that includes only the 'classes' field."""
ds = AttrDict()
classes = [
'__background__', 'articulated_truck', 'bicycle', 'bus', 'car',
'motorcycle', 'motorized_vehicle', 'non-motorized_vehicle',
'pedestrian', 'pickup_truck', 'single_unit_truck light', 'work_van'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_voc_dataset():
"""A dummy COCO dataset that includes only the 'classes' field."""
ds = AttrDict()
classes = [
'__background__', 'ignored_regions', 'pedestrian', 'people', 'bicycle',
'car', 'van', 'truck', ' tricycle', 'awning-tricycle', 'bus', 'motor',
' others'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_vg_dataset():
ds = AttrDict()
obj_cls_file = '/private/home/tinayujiang/data/VisualGenome1.2_in_COCO_format/data_from_peter_anderson/1600-400-20/objects_vocab.txt'
with open(obj_cls_file) as f:
data = f.readlines()
classes = ['__background__']
classes.extend([i.strip() for i in data])
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_voc2007_dataset():
"""A dummy COCO dataset that includes only the 'classes' field."""
ds = AttrDict()
classes = [
'__background__', 'aeroplane', 'bicycle', 'bird', 'boat', 'bottle',
'bus', 'car', 'cat', 'chair', 'cow', 'diningtable', 'dog', 'horse',
'motorbike', 'person', 'pottedplant', 'sheep', 'sofa', 'train',
'tvmonitor'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_custom_dummy_dataset(annFile):
"""A dummy classes generator from coco json file."""
ds = AttrDict()
coco = COCO(annFile)
category_ids = coco.getCatIds()
categories = [c['name'] for c in coco.loadCats(category_ids)]
category_to_id_map = dict(zip(categories, category_ids))
classes = ['__background__'] + categories
print(classes)
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_apollo_laneseg_dataset():
ds = AttrDict()
classes = [
'__background__', 's_w_d', 's_y_d', 'ds_w_dn', 'ds_y_dn', 'sb_w_do',
'sb_y_do', 'b_w_g', 'b_y_g', 'db_w_g', 'db_y_g', 'db_w_s', 's_w_s',
'ds_w_s', 's_w_c', 's_y_c', 's_w_p', 's_n_p', 'c_wy_z', 'a_w_u',
'a_w_t', 'a_w_tl', 'a_w_tr', 'a_w_tlr', 'a_w_l', 'a_w_r', 'a_w_lr',
'a_n_lu', 'a_w_tu', 'a_w_m', 'a_y_t', 'b_n_sr', 'd_wy_za', 'r_wy_np',
'vom_wy_n', 'om_n_n'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_abu_musa_dataset():
"""A dummy COCO dataset that includes only the 'classes' field."""
ds = AttrDict()
classes = [
'__background__', 'Underground Shelter', 'Communication Tower',
'Dense Structures', 'Vehicle', 'Cargo container', 'Ship',
'Swimming pool', 'Sports field', 'Storage Tank', 'Standalone Building',
'Defensive Earthworks'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def get_steal_oil_class10_dataset():
"""A dummy COCO dataset that includes only the 'classes' field."""
ds = AttrDict()
classes = [
'__background__', 'autotruck', 'forklift', 'digger', 'car', 'bus',
'tanker', 'person', 'minitruck', 'minibus'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
# [{"supercategory": "none", "id": 1, "name": "autotruck"}, {"supercategory": "none", "id": 2, "name": "forklift"}, {"supercategory": "none", "id": 3, "name": "digger"}, {"supercategory": "none", "id": 4, "name": "car"}, {"supercategory": "none", "id": 5, "name": "bus"}, {"supercategory": "none", "id": 6, "name": "tanker"}, {"supercategory": "none", "id": 7, "name": "person"}]
def get_cityscape_dataset():
"""A dummy cityscape dataset that includes only the 'classes' field."""
ds = AttrDict()
ds.classes = {i: name for i, name in enumerate(classes)}
ds.classes_name_to_num = {name: i for i, name in enumerate(classes)}
ds.coco_to_this = {}
for label in labels:
name = label.name
if name in classes and label.hasInstances and not label.ignoreInEval:
ds.coco_to_this[ds.classes_name_to_num[name]] = label.id
ds.confident_threshold = 0.5
return ds
def get_mapillary_dataset():
ds = AttrDict()
classes = [
'__background__', 'Bird', 'Ground Animal', 'Crosswalk - Plain',
'Person', 'Bicyclist', 'Motorcyclist', 'Other Rider',
'Lane Marking - Crosswalk', 'Banner', 'Bench', 'Bike Rack',
'Billboard', 'Catch Basin', 'CCTV Camera', 'Fire Hydrant',
'Junction Box', 'Mailbox', 'Manhole', 'Phone Booth', 'Street Light',
'Pole', 'Traffic Sign Frame', 'Utility Pole', 'Traffic Light',
'Traffic Sign (Back)', 'Traffic Sign (Front)', 'Trash Can', 'Bicycle',
'Boat', 'Bus', 'Car', 'Caravan', 'Motorcycle', 'Other Vehicle',
'Trailer', 'Truck', 'Wheeled Slow'
]
ds.classes = {i: name for i, name in enumerate(classes)}
return ds
def test_merge_cfg_from_cfg(self):
# Test: merge from deepcopy
s = 'dummy0'
cfg2 = copy.deepcopy(cfg)
cfg2.MODEL.TYPE = s
core.config.merge_cfg_from_cfg(cfg2)
assert cfg.MODEL.TYPE == s
# Test: merge from yaml
s = 'dummy1'
cfg2 = yaml.load(yaml.dump(cfg))
cfg2.MODEL.TYPE = s
core.config.merge_cfg_from_cfg(cfg2)
assert cfg.MODEL.TYPE == s
# Test: merge with a valid key
s = 'dummy2'
cfg2 = AttrDict()
cfg2.MODEL = AttrDict()
cfg2.MODEL.TYPE = s
core.config.merge_cfg_from_cfg(cfg2)
assert cfg.MODEL.TYPE == s
# Test: merge with an invalid key
s = 'dummy3'
cfg2 = AttrDict()
cfg2.FOO = AttrDict()
cfg2.FOO.BAR = s
with self.assertRaises(KeyError):
core.config.merge_cfg_from_cfg(cfg2)
# Test: merge with converted type
cfg2 = AttrDict()
cfg2.TRAIN = AttrDict()
cfg2.TRAIN.SCALES = [1]
core.config.merge_cfg_from_cfg(cfg2)
assert type(cfg.TRAIN.SCALES) is tuple
assert cfg.TRAIN.SCALES[0] == 1
# Test: merge with invalid type
cfg2 = AttrDict()
cfg2.TRAIN = AttrDict()
cfg2.TRAIN.SCALES = 1
with self.assertRaises(ValueError):
core.config.merge_cfg_from_cfg(cfg2)
def _merge_a_into_b(a, b):
"""Merge config dictionary a into config dictionary b, clobbering the
options in b whenever they are also specified in a.
"""
from ast import literal_eval
if not isinstance(a, AttrDict):
return
for k, v in a.items():
# a must specify keys that are in b
if k not in b:
if k + '_deprecated' in b:
logger.warn('Config key {} is deprecated, ignoring'.format(k))
return
else:
raise KeyError('{} is not a valid config key'.format(k))
if type(v) is dict:
a[k] = v = AttrDict(v)
if isinstance(v, basestring): # NoQA
try:
v = literal_eval(v)
except BaseException:
pass
# the types must match, too (with some exceptions)
old_type = type(b[k])
if old_type is not type(v) and v is not None:
if isinstance(b[k], np.ndarray):
v = np.array(v, dtype=b[k].dtype)
elif isinstance(b[k], basestring) and isinstance(v, unicode): # NoQA
v = str(v)
else:
raise ValueError(
'Type mismatch ({} vs. {}) for config key: {}'.format(
type(b[k]), type(v), k))
# recursively merge dicts
if isinstance(v, AttrDict):
try:
_merge_a_into_b(a[k], b[k])
except BaseException:
logger.critical('Error under config key: {}'.format(k))
raise
else:
b[k] = v
def get_quality_dataset():
ds = AttrDict()
classes = [
'偏左', '偏右', '偏上', '偏下', '左肩胛骨在肺野内', '右肩胛骨在肺野内', '异物', '标记在软组织或锁骨上',
'栅切割伪影', '心影后脊柱不清', '双肺纹理模糊'
]
cls2id = {name: i for i, name in enumerate(classes)}
# chi2eng = {'肺实变': 'consolidation', '纤维化表现': 'fibrosis', '肋骨异常': 'rib_abnormity', '胸腔积液': 'effusion',
# '胸膜增厚': 'pleural_thickening', '主动脉异常': 'aorta_abnormity', '膈面异常': 'diaphragm_abnormity',
# '膈下游离气体': 'subphrenic_air', '结节': 'nodule', '肿块': 'mass', '异物': 'foreign_matters',
# '气胸': 'pneumothorax', '肺气肿': 'emphysema', '骨折': 'rib_fracture'}
#
# cls2chi = {cls: chi for cls, chi in enumerate(classes)}
# cls2eng = {cls: chi2eng[chi] for cls, chi in enumerate(classes)}
th_cls = [
0,
0.1,
0.1,
0.1,
0.1,
0.1,
0.1,
0.1,
0.1,
0.1, # threshold for each cls
0.1,
0.1,
0.1,
0.1,
0.1,
0.1,
0.1,
0.1,
0.1,
0.1
]
# ds.classes = {i: name for i, name in enumerate(classes)}
ds.classes = classes
ds.th_cls = th_cls
ds.cls2id = cls2id
return ds
def merge_cfg_from_file(cfg_filename):
"""Load a yaml config file and merge it into the global config."""
with open(cfg_filename, 'r') as f:
yaml_cfg = AttrDict(yaml.load(f, Loader=yaml.FullLoader))
_merge_a_into_b(yaml_cfg, __C)
import os import os.path as osp import copy from ast import literal_eval import numpy as np from packaging import version import torch import torch.nn as nn from torch.nn import init import yaml import nn as mynn from utils.collections import AttrDict __C = AttrDict() # Consumers can get config by: # from fast_rcnn_config import cfg cfg = __C # Random note: avoid using '.ON' as a config key since yaml converts it to True; # prefer 'ENABLED' instead # ---------------------------------------------------------------------------- # # Training options # ---------------------------------------------------------------------------- # __C.TRAIN = AttrDict() # Datasets to train on # Available dataset list: datasets.dataset_catalog.DATASETS.keys() # If multiple datasets are listed, the model is trained on their union
