def image_compute_signatures(img_path):
# compute chain codes for all fragments
chaincodes = chaincode.chaincodes(img_path)
documents = text_embedding.chaincodes_to_documents(chaincodes)
signatures = jaccard.document_signatures(documents)
return signatures, chaincodes
def test_minhash_chaincodes(self):
QUERY_INDEX = 3
# compute chain codes for all fragments
chaincodes = chaincode.chaincodes(self.img_path)
# find the chain code for query chain
q_code = text_embedding.query_chaincode(chaincodes[QUERY_INDEX])
documents = text_embedding.chaincodes_to_documents(chaincodes)
q_document = text_embedding.chaincodes_to_documents([q_code])
signatures = jaccard.document_signatures(documents)
q_signature = jaccard.document_signatures(q_document)[0]
lsh_indices = jaccard.create_buckets(signatures)
similar_documents = jaccard.query(lsh_indices, q_signature, QUERY_INDEX)
print(f'similar_documents: {similar_documents}')
self.assertGreater(len(similar_documents), 0)
def image_query_similar_fragments(chaincodes, signatures, query_index,
threshold):
lsh_indices = jaccard.create_buckets(signatures, threshold)
# find the chain code for query chain
q_code = text_embedding.query_chaincode(chaincodes[query_index])
q_document = text_embedding.chaincodes_to_documents([q_code])
q_signature = jaccard.document_signatures(q_document)[0]
similar_documents = jaccard.query(lsh_indices, q_signature, query_index,
threshold)
return similar_documents
def test_can_get_similar_documents(self):
QUERY_INDEX = 1
signatures = jaccard.document_signatures(self.documents)
lsh_indices = jaccard.create_buckets(signatures)
similar_documents = jaccard.query(lsh_indices, signatures[QUERY_INDEX],
QUERY_INDEX)
def test_can_get_signatures_for_multiple_documents(self):
signatures = jaccard.document_signatures(self.documents)
self.assertIsInstance(signatures, list)
self.assertIsInstance(signatures[0], MinHash)