Add `mmax` config key to auto_augment for increasing upper bound of RandAugment magnitude beyond 10. Make AugMix uniform sampling default not override config setting.

pull/805/head
Ross Wightman 3 years ago
parent 1042b8a146
commit 3cdaf5ed56

@ -29,9 +29,7 @@ _PIL_VER = tuple([int(x) for x in PIL.__version__.split('.')[:2]])
_FILL = (128, 128, 128) _FILL = (128, 128, 128)
# This signifies the max integer that the controller RNN could predict for the _LEVEL_DENOM = 10. # denominator for conversion from 'Mx' magnitude scale to fractional aug level for op arguments
# augmentation scheme.
_MAX_LEVEL = 10.
_HPARAMS_DEFAULT = dict( _HPARAMS_DEFAULT = dict(
translate_const=250, translate_const=250,
@ -179,34 +177,34 @@ def _randomly_negate(v):
def _rotate_level_to_arg(level, _hparams): def _rotate_level_to_arg(level, _hparams):
# range [-30, 30] # range [-30, 30]
level = (level / _MAX_LEVEL) * 30. level = (level / _LEVEL_DENOM) * 30.
level = _randomly_negate(level) level = _randomly_negate(level)
return level, return level,
def _enhance_level_to_arg(level, _hparams): def _enhance_level_to_arg(level, _hparams):
# range [0.1, 1.9] # range [0.1, 1.9]
return (level / _MAX_LEVEL) * 1.8 + 0.1, return (level / _LEVEL_DENOM) * 1.8 + 0.1,
def _enhance_increasing_level_to_arg(level, _hparams): def _enhance_increasing_level_to_arg(level, _hparams):
# the 'no change' level is 1.0, moving away from that towards 0. or 2.0 increases the enhancement blend # the 'no change' level is 1.0, moving away from that towards 0. or 2.0 increases the enhancement blend
# range [0.1, 1.9] # range [0.1, 1.9] if level <= _LEVEL_DENOM
level = (level / _MAX_LEVEL) * .9 level = (level / _LEVEL_DENOM) * .9
level = 1.0 + _randomly_negate(level) level = max(0.1, 1.0 + _randomly_negate(level)) # keep it >= 0.1
return level, return level,
def _shear_level_to_arg(level, _hparams): def _shear_level_to_arg(level, _hparams):
# range [-0.3, 0.3] # range [-0.3, 0.3]
level = (level / _MAX_LEVEL) * 0.3 level = (level / _LEVEL_DENOM) * 0.3
level = _randomly_negate(level) level = _randomly_negate(level)
return level, return level,
def _translate_abs_level_to_arg(level, hparams): def _translate_abs_level_to_arg(level, hparams):
translate_const = hparams['translate_const'] translate_const = hparams['translate_const']
level = (level / _MAX_LEVEL) * float(translate_const) level = (level / _LEVEL_DENOM) * float(translate_const)
level = _randomly_negate(level) level = _randomly_negate(level)
return level, return level,
@ -214,7 +212,7 @@ def _translate_abs_level_to_arg(level, hparams):
def _translate_rel_level_to_arg(level, hparams): def _translate_rel_level_to_arg(level, hparams):
# default range [-0.45, 0.45] # default range [-0.45, 0.45]
translate_pct = hparams.get('translate_pct', 0.45) translate_pct = hparams.get('translate_pct', 0.45)
level = (level / _MAX_LEVEL) * translate_pct level = (level / _LEVEL_DENOM) * translate_pct
level = _randomly_negate(level) level = _randomly_negate(level)
return level, return level,
@ -223,7 +221,7 @@ def _posterize_level_to_arg(level, _hparams):
# As per Tensorflow TPU EfficientNet impl # As per Tensorflow TPU EfficientNet impl
# range [0, 4], 'keep 0 up to 4 MSB of original image' # range [0, 4], 'keep 0 up to 4 MSB of original image'
# intensity/severity of augmentation decreases with level # intensity/severity of augmentation decreases with level
return int((level / _MAX_LEVEL) * 4), return int((level / _LEVEL_DENOM) * 4),
def _posterize_increasing_level_to_arg(level, hparams): def _posterize_increasing_level_to_arg(level, hparams):
@ -237,13 +235,13 @@ def _posterize_original_level_to_arg(level, _hparams):
# As per original AutoAugment paper description # As per original AutoAugment paper description
# range [4, 8], 'keep 4 up to 8 MSB of image' # range [4, 8], 'keep 4 up to 8 MSB of image'
# intensity/severity of augmentation decreases with level # intensity/severity of augmentation decreases with level
return int((level / _MAX_LEVEL) * 4) + 4, return int((level / _LEVEL_DENOM) * 4) + 4,
def _solarize_level_to_arg(level, _hparams): def _solarize_level_to_arg(level, _hparams):
# range [0, 256] # range [0, 256]
# intensity/severity of augmentation decreases with level # intensity/severity of augmentation decreases with level
return int((level / _MAX_LEVEL) * 256), return int((level / _LEVEL_DENOM) * 256),
def _solarize_increasing_level_to_arg(level, _hparams): def _solarize_increasing_level_to_arg(level, _hparams):
@ -254,7 +252,7 @@ def _solarize_increasing_level_to_arg(level, _hparams):
def _solarize_add_level_to_arg(level, _hparams): def _solarize_add_level_to_arg(level, _hparams):
# range [0, 110] # range [0, 110]
return int((level / _MAX_LEVEL) * 110), return int((level / _LEVEL_DENOM) * 110),
LEVEL_TO_ARG = { LEVEL_TO_ARG = {
@ -334,17 +332,22 @@ class AugmentOp:
# NOTE This is my own hack, being tested, not in papers or reference impls. # NOTE This is my own hack, being tested, not in papers or reference impls.
# If magnitude_std is inf, we sample magnitude from a uniform distribution # If magnitude_std is inf, we sample magnitude from a uniform distribution
self.magnitude_std = self.hparams.get('magnitude_std', 0) self.magnitude_std = self.hparams.get('magnitude_std', 0)
self.magnitude_max = self.hparams.get('magnitude_max', None)
def __call__(self, img): def __call__(self, img):
if self.prob < 1.0 and random.random() > self.prob: if self.prob < 1.0 and random.random() > self.prob:
return img return img
magnitude = self.magnitude magnitude = self.magnitude
if self.magnitude_std: if self.magnitude_std > 0:
# magnitude randomization enabled
if self.magnitude_std == float('inf'): if self.magnitude_std == float('inf'):
magnitude = random.uniform(0, magnitude) magnitude = random.uniform(0, magnitude)
elif self.magnitude_std > 0: elif self.magnitude_std > 0:
magnitude = random.gauss(magnitude, self.magnitude_std) magnitude = random.gauss(magnitude, self.magnitude_std)
magnitude = min(_MAX_LEVEL, max(0, magnitude)) # clip to valid range # default upper_bound for the timm RA impl is _LEVEL_DENOM (10)
# setting magnitude_max overrides this to allow M > 10 (behaviour closer to Google TF RA impl)
upper_bound = self.magnitude_max or _LEVEL_DENOM
magnitude = max(0., min(magnitude, upper_bound))
level_args = self.level_fn(magnitude, self.hparams) if self.level_fn is not None else tuple() level_args = self.level_fn(magnitude, self.hparams) if self.level_fn is not None else tuple()
return self.aug_fn(img, *level_args, **self.kwargs) return self.aug_fn(img, *level_args, **self.kwargs)
@ -642,7 +645,8 @@ def rand_augment_transform(config_str, hparams):
'm' - integer magnitude of rand augment 'm' - integer magnitude of rand augment
'n' - integer num layers (number of transform ops selected per image) 'n' - integer num layers (number of transform ops selected per image)
'w' - integer probabiliy weight index (index of a set of weights to influence choice of op) 'w' - integer probabiliy weight index (index of a set of weights to influence choice of op)
'mstd' - float std deviation of magnitude noise applied 'mstd' - float std deviation of magnitude noise applied, or uniform sampling if infinity (or > 100)
'mmax' - set upper bound for magnitude to something other than default of _LEVEL_DENOM (10)
'inc' - integer (bool), use augmentations that increase in severity with magnitude (default: 0) 'inc' - integer (bool), use augmentations that increase in severity with magnitude (default: 0)
Ex 'rand-m9-n3-mstd0.5' results in RandAugment with magnitude 9, num_layers 3, magnitude_std 0.5 Ex 'rand-m9-n3-mstd0.5' results in RandAugment with magnitude 9, num_layers 3, magnitude_std 0.5
'rand-mstd1-w0' results in magnitude_std 1.0, weights 0, default magnitude of 10 and num_layers 2 'rand-mstd1-w0' results in magnitude_std 1.0, weights 0, default magnitude of 10 and num_layers 2
@ -651,7 +655,7 @@ def rand_augment_transform(config_str, hparams):
:return: A PyTorch compatible Transform :return: A PyTorch compatible Transform
""" """
magnitude = _MAX_LEVEL # default to _MAX_LEVEL for magnitude (currently 10) magnitude = _LEVEL_DENOM # default to _LEVEL_DENOM for magnitude (currently 10)
num_layers = 2 # default to 2 ops per image num_layers = 2 # default to 2 ops per image
weight_idx = None # default to no probability weights for op choice weight_idx = None # default to no probability weights for op choice
transforms = _RAND_TRANSFORMS transforms = _RAND_TRANSFORMS
@ -664,8 +668,15 @@ def rand_augment_transform(config_str, hparams):
continue continue
key, val = cs[:2] key, val = cs[:2]
if key == 'mstd': if key == 'mstd':
# noise param injected via hparams for now # noise param / randomization of magnitude values
hparams.setdefault('magnitude_std', float(val)) mstd = float(val)
if mstd > 100:
# use uniform sampling in 0 to magnitude if mstd is > 100
mstd = float('inf')
hparams.setdefault('magnitude_std', mstd)
elif key == 'mmax':
# clip magnitude between [0, mmax] instead of default [0, _LEVEL_DENOM]
hparams.setdefault('magnitude_max', int(val))
elif key == 'inc': elif key == 'inc':
if bool(val): if bool(val):
transforms = _RAND_INCREASING_TRANSFORMS transforms = _RAND_INCREASING_TRANSFORMS
@ -794,7 +805,6 @@ def augment_and_mix_transform(config_str, hparams):
depth = -1 depth = -1
alpha = 1. alpha = 1.
blended = False blended = False
hparams['magnitude_std'] = float('inf')
config = config_str.split('-') config = config_str.split('-')
assert config[0] == 'augmix' assert config[0] == 'augmix'
config = config[1:] config = config[1:]
@ -818,5 +828,6 @@ def augment_and_mix_transform(config_str, hparams):
blended = bool(val) blended = bool(val)
else: else:
assert False, 'Unknown AugMix config section' assert False, 'Unknown AugMix config section'
hparams.setdefault('magnitude_std', float('inf')) # default to uniform sampling (if not set via mstd arg)
ops = augmix_ops(magnitude=magnitude, hparams=hparams) ops = augmix_ops(magnitude=magnitude, hparams=hparams)
return AugMixAugment(ops, alpha=alpha, width=width, depth=depth, blended=blended) return AugMixAugment(ops, alpha=alpha, width=width, depth=depth, blended=blended)

Loading…
Cancel
Save