def build_flann_index(features, config):
FLANN_INDEX_LINEAR = 0
FLANN_INDEX_KDTREE = 1
FLANN_INDEX_KMEANS = 2
FLANN_INDEX_COMPOSITE = 3
FLANN_INDEX_KDTREE_SINGLE = 4
FLANN_INDEX_HIERARCHICAL = 5
FLANN_INDEX_LSH = 6
if features.dtype.type is np.float32:
algorithm_type = config['flann_algorithm'].upper()
if algorithm_type == 'KMEANS':
FLANN_INDEX_METHOD = FLANN_INDEX_KMEANS
elif algorithm_type == 'KDTREE':
FLANN_INDEX_METHOD = FLANN_INDEX_KDTREE
else:
raise ValueError('Unknown flann algorithm type '
'must be KMEANS, KDTREE')
else:
FLANN_INDEX_METHOD = FLANN_INDEX_LSH
flann_params = dict(algorithm=FLANN_INDEX_METHOD,
branching=config['flann_branching'],
iterations=config['flann_iterations'],
tree=config['flann_tree'])
return context.flann_Index(features, flann_params)
def build_flann_index(features: np.ndarray, config: Dict[str, Any]) -> Any:
# FLANN_INDEX_LINEAR = 0
FLANN_INDEX_KDTREE = 1
FLANN_INDEX_KMEANS = 2
# FLANN_INDEX_COMPOSITE = 3
# FLANN_INDEX_KDTREE_SINGLE = 4
# FLANN_INDEX_HIERARCHICAL = 5
FLANN_INDEX_LSH = 6
if features.dtype.type is np.float32:
algorithm_type = config["flann_algorithm"].upper()
if algorithm_type == "KMEANS":
FLANN_INDEX_METHOD = FLANN_INDEX_KMEANS
elif algorithm_type == "KDTREE":
FLANN_INDEX_METHOD = FLANN_INDEX_KDTREE
else:
raise ValueError("Unknown flann algorithm type "
"must be KMEANS, KDTREE")
else:
FLANN_INDEX_METHOD = FLANN_INDEX_LSH
flann_params = {
"algorithm": FLANN_INDEX_METHOD,
"branching": config["flann_branching"],
"iterations": config["flann_iterations"],
"tree": config["flann_tree"],
}
return context.flann_Index(features, flann_params)
def __init__(self, words, frequencies):
self.words = words
self.frequencies = frequencies
self.weights = np.log(frequencies.sum() / frequencies)
FLANN_INDEX_KDTREE = 1
flann_params = {"algorithm": FLANN_INDEX_KDTREE, "trees": 8, "checks": 300}
self.index = context.flann_Index(words, flann_params)
def load_feature_index(self, image, features):
if not self.feature_index_exists(image):
raise IOError(
"FLANN index file for {} doesn't exists.".format(image))
index = context.flann_Index()
index.load(features, self._feature_index_file(image))
return index
def __init__(self, words, frequencies):
self.words = words
self.frequencies = frequencies
self.weights = np.log(frequencies.sum() / frequencies)
FLANN_INDEX_KDTREE = 1
flann_params = dict(algorithm=FLANN_INDEX_KDTREE, trees=8, checks=300)
self.index = context.flann_Index(words, flann_params)
def build_flann_index(descriptors: np.ndarray, config: Dict[str, Any]) -> Any:
# FLANN_INDEX_LINEAR = 0
FLANN_INDEX_KDTREE = 1
FLANN_INDEX_KMEANS = 2
# FLANN_INDEX_COMPOSITE = 3
# FLANN_INDEX_KDTREE_SINGLE = 4
# FLANN_INDEX_HIERARCHICAL = 5
if descriptors.dtype.type is np.float32:
algorithm_type = config["flann_algorithm"].upper()
if algorithm_type == "KMEANS":
FLANN_INDEX_METHOD = FLANN_INDEX_KMEANS
elif algorithm_type == "KDTREE":
FLANN_INDEX_METHOD = FLANN_INDEX_KDTREE
else:
raise ValueError("Unknown flann algorithm type "
"must be KMEANS, KDTREE")
flann_params = {
"algorithm": FLANN_INDEX_METHOD,
"branching": config["flann_branching"],
"iterations": config["flann_iterations"],
"tree": config["flann_tree"],
}
else:
raise ValueError(
"FLANN isn't supported for binary features because of poor-performance. Use BRUTEFORCE instead."
)
return context.flann_Index(descriptors, flann_params)
def build_flann_index(features, config):
FLANN_INDEX_LINEAR = 0
FLANN_INDEX_KDTREE = 1
FLANN_INDEX_KMEANS = 2
FLANN_INDEX_COMPOSITE = 3
FLANN_INDEX_KDTREE_SINGLE = 4
FLANN_INDEX_HIERARCHICAL = 5
FLANN_INDEX_LSH = 6
if features.dtype.type is np.float32:
FLANN_INDEX_METHOD = FLANN_INDEX_KMEANS
else:
FLANN_INDEX_METHOD = FLANN_INDEX_LSH
flann_params = dict(algorithm=FLANN_INDEX_METHOD,
branching=config['flann_branching'],
iterations=config['flann_iterations'])
return context.flann_Index(features, flann_params)
def build_flann_index(features, config):
FLANN_INDEX_LINEAR = 0
FLANN_INDEX_KDTREE = 1
FLANN_INDEX_KMEANS = 2
FLANN_INDEX_COMPOSITE = 3
FLANN_INDEX_KDTREE_SINGLE = 4
FLANN_INDEX_HIERARCHICAL = 5
FLANN_INDEX_LSH = 6
if features.dtype.type is np.float32:
FLANN_INDEX_METHOD = FLANN_INDEX_KMEANS
else:
FLANN_INDEX_METHOD = FLANN_INDEX_LSH
flann_params = dict(algorithm=FLANN_INDEX_METHOD,
branching=config['flann_branching'],
iterations=config['flann_iterations'])
return context.flann_Index(features, flann_params)
def load_feature_index(self, image, features):
index = context.flann_Index()
index.load(features, self._feature_index_file(image))
return index
def load_feature_index(self, image, features):
index = context.flann_Index()
index.load(features, self._feature_index_file(image))
return index
