_C.resize_width = 640 _C.resize_height = 640 _C.scale = 1 / 127.0 _C.anchor_sampling = True _C.filter_min_face = True # train config _C.LR_STEPS = (80000, 100000, 120000) _C.MAX_STEPS = 150000 _C.EPOCHES = 100 # anchor config _C.FEATURE_MAPS = [160, 80, 40, 20, 10, 5] _C.INPUT_SIZE = 640 _C.STEPS = [4, 8, 16, 32, 64, 128] _C.ANCHOR_SIZES = [16, 32, 64, 128, 256, 512] _C.CLIP = False _C.VARIANCE = [0.1, 0.2] # loss config _C.NUM_CLASSES = 2 _C.OVERLAP_THRESH = 0.35 _C.NEG_POS_RATIOS = 3 # detection config _C.NMS_THRESH = 0.3 _C.TOP_K = 5000 _C.KEEP_TOP_K = 750 _C.CONF_THRESH = 0.05 # dataset config
_C.LR_STEPS = (80000,100000,120000) _C.EPOCHES = 300 # anchor config """ _C.FEATURE_MAPS = [160, 80, 40, 20, 10, 5] _C.INPUT_SIZE = 640 _C.STEPS = [4, 8, 16, 32, 64, 128] _C.ANCHOR_SIZES = [16, 32, 64, 128, 256, 512] _C.CLIP = False _C.VARIANCE = [0.1, 0.2] """ _C.FEATURE_MAPS = [5, 10, 20, 40, 80, 160] _C.INPUT_SIZE = 640 _C.STEPS = [128, 64, 32, 16, 8, 4] _C.ANCHOR_SIZES = [512, 256, 128, 64, 32, 16] _C.CLIP = False _C.VARIANCE = [0.1, 0.2] # detection config _C.NMS_THRESH = 0.3 _C.NMS_TOP_K = 5000 _C.TOP_K = 750 _C.CONF_THRESH = 0.05 # loss config _C.NEG_POS_RATIOS = 3 _C.NUM_CLASSES = 2 _C.USE_NMS = True # dataset config
_C.apply_expand = False
_C.img_mean = np.array([104., 117., 123.])[:, np.newaxis,
np.newaxis].astype('float32')
_C.resize_width = 320
_C.resize_height = 320
_C.scale = 1 / 127.0
_C.anchor_sampling = True
_C.filter_min_face = True
_C.IS_MONOCHROME = True
# anchor config
_C.FEATURE_MAPS = [40, 40, 20, 20]
_C.INPUT_SIZE = 320
_C.STEPS = [8, 8, 16, 16]
_C.ANCHOR_SIZES = [8, 16, 32, 48]
_C.CLIP = False
_C.VARIANCE = [0.1, 0.2]
# detection config
_C.NMS_THRESH = 0.3
_C.NMS_TOP_K = 5000
_C.TOP_K = 750
_C.CONF_THRESH = 0.05
# loss config
_C.NEG_POS_RATIOS = 3
_C.NUM_CLASSES = 2
_C.USE_NMS = True
# face config
_C.FDDB_DIR = '/home/lj/data/FDDB' _C.WIDER_DIR = '/home/lj/data/WIDER' _C.AFW_DIR = '/home/lj/data/AFW' _C.PASCAL_DIR = '/home/lj/data/PASCAL_FACE' # train config _C.MAX_STEPS = 120000 _C.LR_STEPS = (80000,100000,120000) _C.EPOCHES = 300 # anchor config _C.FEATURE_MAPS = [32, 16, 8] _C.STEPS = [32,64,128] _C.DENSITY = [[-3, -1, 1, 3], [-2, 2], [0]] _C.ASPECT_RATIOS = ((1, 2, 4), (1,), (1,)) _C.ANCHOR_SIZES = [32, 256, 512] _C.VARIANCE = [0.1, 0.2] _C.CLIP = False # loss config _C.NUM_CLASSES = 2 _C.OVERLAP_THRESH = 0.35 _C.NEG_POS_RATION = 7 # detection config _C.NMS_THRESH = 0.3 _C.NMS_TOP_K = 5000 _C.KEEP_TOP_K = 750 _C.CONF_THRESH = 0.05
import math
from easydict import EasyDict
__C = EasyDict()
cfg = __C
__C.DATASET_NAME = ''
__C.INPUT_SIZE = (600, 600)
__C.CLASS_NUM = 20
__C.CLASSES = ('aeroplane', 'bicycle', 'bird', 'boat', 'bottle', 'bus', 'car',
'cat', 'chair', 'cow', 'diningtable', 'dog', 'horse',
'motorbike', 'person', 'pottedplant', 'sheep', 'sofa', 'train',
'tvmonitor')
__C.ANCHOR_SIZES = [32, 64, 128, 256, 512]
__C.ASPECT_RATIOS = [0.5, 1.0, 2.0]
__C.SCALE_RATIOS = [pow(2, 0 / 3.), pow(2, 1 / 3.), pow(2, 2 / 3.)]
