def __init__(self):
Indexer.__init__(self)
self.encoder = SHEncoder()
self.set_storage()
class SHIndexer(Indexer):
def __init__(self):
Indexer.__init__(self)
self.encoder = SHEncoder()
self.set_storage()
def __del__(self):
pass
def build(self, pardic=None):
self.encoder.build(pardic)
def set_storage(self, storage_type='mem', storage_parm=None):
self.storage = createStorage(storage_type, storage_parm)
def add(self, vals, keys=None):
num_vals = vals.shape[0]
if keys is None:
num_base_items = self.storage.get_num_items()
keys = np.arange(num_base_items, num_base_items + num_vals,
dtype=np.int32)
else:
keys = np.array(keys, dtype=np.int32).reshape(-1)
start_id = 0
for start_id in range(0, num_vals, self.BLKSIZE):
cur_num = min(self.BLKSIZE, num_vals - start_id)
logging.info("%8d/%d: %d" % (start_id, num_vals, cur_num))
codes = self.encoder.encode(vals[start_id:start_id+cur_num, :])
self.storage.add(codes, keys[start_id:start_id+cur_num])
def remove(self, keys):
raise Exception(self.ERR_UNIMPL)
@staticmethod
def hammingDist(B1, B2):
"""
Compute hamming distance between two sets of samples (B1, B2)
Dh=hammingDist(B1, B2);
Input
B1, B2: compact bit vectors. Each datapoint is one row.
size(B1) = [ndatapoints1, nwords]
size(B2) = [ndatapoints2, nwords]
It is faster if ndatapoints1 < ndatapoints2
Output
Dh = hamming distance.
size(Dh) = [ndatapoints1, ndatapoints2]
example query
Dhamm = hammingDist(B2, B1);
this will give the same result than:
Dhamm = distMat(U2>0, U1>0).^2;
the size of the distance matrix is:
size(Dhamm) = [Ntest x Ntraining]
"""
if B1.ndim == 1:
B1 = B1.reshape((1, -1))
if B2.ndim == 1:
B2 = B2.reshape((1, -1))
npt1, dim1 = B1.shape
npt2, dim2 = B2.shape
if dim1 != dim2:
raise Exception("Dimension not consists: %d, %d" % (dim1, dim2))
Dh = np.zeros((npt1, npt2), np.uint16)
for i in xrange(npt1):
Dh[i, :] = BIT_CNT_MAP[np.bitwise_xor(B1[i, :], B2)].sum(1)
return Dh
@staticmethod
def hammingDist2(B1, B2):
"""
Compute hamming distance between two sets of samples (B1, B2)
Dh=hammingDist(B1, B2);
Input
B1, B2: compact bit vectors. Each datapoint is one row.
size(B1) = [ndatapoints1, nwords]
size(B2) = [ndatapoints2, nwords]
It is faster if ndatapoints1 < ndatapoints2
Output
Dh = hamming distance.
size(Dh) = [ndatapoints1, ndatapoints2]
example query
Dhamm = hammingDist(B2, B1);
this will give the same result than:
Dhamm = distMat(U2>0, U1>0).^2;
the size of the distance matrix is:
size(Dhamm) = [Ntest x Ntraining]
"""
if B1.ndim == 1:
B1 = B1.reshape((1, -1))
if B2.ndim == 1:
B2 = B2.reshape((1, -1))
npt1, dim1 = B1.shape
npt2, dim2 = B2.shape
if dim1 != dim2:
raise Exception("Dimension not consists: %d, %d" % (dim1, dim2))
Dh = cext.hamming(B1, B2)
return Dh
def search(self, queries, topk=None, **kwargs):
nq = queries.shape[0]
nbits = self.encoder.ecdat['nbits']
# qry_codes = self.encoder.encode(queries)
db_codes = self.storage.get_codes()
idsquerybase = self.storage.get_keys()
dis = np.ones((nq, topk), np.single) * np.inf
ids = np.ones((nq, topk), np.int32) * -1
profiler = Profiler()
interval = 100 if nq >= 100 else 10
time_total = 0.0 # total time for all queries
logging.info('Start Querying ...')
for qry_id in range(nq):
profiler.start("encoding") # time for computing the distances
qry_code = self.encoder.encode(queries[qry_id:qry_id+1])
profiler.end()
profiler.start("distance") # time for computing the distances
disquerybase = self.hammingDist2(qry_code, db_codes).reshape(-1)
profiler.end()
profiler.start("knn") # time for finding the kNN
cur_ids = cext.knn_count(disquerybase, nbits, topk)
profiler.end()
profiler.start("result") # time for getting final result
ids[qry_id, :] = idsquerybase[cur_ids]
dis[qry_id, :] = disquerybase[cur_ids]
profiler.end()
if (qry_id+1) % interval == 0:
time_total += profiler.sum_overall()
logging.info(
'\t%d/%d: %.3fms per query' %
(qry_id+1, nq, profiler.sum_average() * 1000))
logging.info("\t\t%s" % profiler.str_average())
profiler.reset()
logging.info('Querying Finished!')
time_total += profiler.sum_overall()
logging.info("Average querying time: %.3fms" % (time_total * 1000 / nq))
return ids, dis
