def test_ctc_output_probability(self):
seq_len_0 = 2
classes = 3
input_prob_matrix_0 = np.asarray(
[[0.4, 0.00000001, 0.6], [0.4, 0.00000001, 0.6]], dtype=np.float32)
input_log_prob_matrix_0 = np.log(input_prob_matrix_0)
inputs = np.array([
input_log_prob_matrix_0[t, :][np.newaxis, :]
for t in range(seq_len_0)
])
seq_lens = np.array([seq_len_0], dtype=np.int32)
th_input = torch.from_numpy(inputs)
th_seq_len = torch.IntTensor(seq_lens)
labels = "AB_"
scorer = ctcdecode.Scorer()
decoder = ctcdecode.CTCBeamDecoder(scorer,
labels,
blank_index=2,
space_index=-1,
top_paths=1,
beam_width=3)
decode_result, scores, decode_len, alignments, char_probs = decoder.decode(
th_input, th_seq_len)
self.assertEqual(decode_len[0][0], 1)
self.assertEqual(
decode_result.numpy()[0, 0, :decode_len[0][0]].tolist(), [0])
self.assertEqual(alignments.numpy()[0, 0, :decode_len[0][0]].tolist(),
[1])
np.testing.assert_almost_equal(scores.numpy(),
np.log(np.array([[0.64]])), 5)
def test_simple_decode_different_blank_idx(self):
aa = torch.FloatTensor(
np.array([[[0.0, 1.0]], [[0.0, 1.0]], [[1.0, 0.0]], [[0.0, 1.0]],
[[0.0, 1.0]]],
dtype=np.float32)).log()
seq_len = torch.IntTensor(np.array([5], dtype=np.int32))
labels = "_A"
scorer = ctcdecode.Scorer()
decoder_nomerge = ctcdecode.CTCBeamDecoder(scorer,
labels,
blank_index=0,
space_index=-1,
top_paths=1,
beam_width=1)
result_nomerge, _, result_nomerge_len, nomerge_alignments, _ = decoder_nomerge.decode(
aa, seq_len)
self.assertEqual(result_nomerge_len[0][0], 2)
self.assertEqual(
result_nomerge.numpy()[0, 0, :result_nomerge_len[0][0]].tolist(),
[1, 1])
def test_ctc_decoder_beam_search(self):
depth = 6
seq_len_0 = 5
input_prob_matrix_0 = np.asarray(
[
[0.30999, 0.309938, 0.0679938, 0.0673362, 0.0708352, 0.173908],
[
0.215136, 0.439699, 0.0370931, 0.0393967, 0.0381581,
0.230517
],
[
0.199959, 0.489485, 0.0233221, 0.0251417, 0.0233289,
0.238763
],
[0.279611, 0.452966, 0.0204795, 0.0209126, 0.0194803, 0.20655],
[0.51286, 0.288951, 0.0243026, 0.0220788, 0.0219297, 0.129878],
# Random entry added in at time=5
[0.155251, 0.164444, 0.173517, 0.176138, 0.169979, 0.160671]
],
dtype=np.float32)
# Take the log
input_log_prob_matrix_0 = np.log(input_prob_matrix_0)
# len max_time_steps array of batch_size x depth matrices
inputs = np.array(
[
input_log_prob_matrix_0[t, :][np.newaxis, :]
for t in range(seq_len_0)
] # Pad to max_time_steps = 8
+ 2 * [np.zeros((1, depth), dtype=np.float32)],
dtype=np.float32)
# batch_size length vector of sequence_lengths
seq_lens = np.array([seq_len_0], dtype=np.int32)
th_input = torch.from_numpy(inputs)
th_seq_len = torch.IntTensor(seq_lens)
labels = "ABCDE_"
scorer = ctcdecode.Scorer()
decoder = ctcdecode.CTCBeamDecoder(scorer,
labels,
blank_index=5,
space_index=-1,
top_paths=2,
beam_width=2)
decode_result, scores, decode_len, alignments, _ = decoder.decode(
th_input, th_seq_len)
self.assertEqual(decode_len[0][0], 2)
self.assertEqual(decode_len[1][0], 3)
self.assertEqual(
decode_result.numpy()[0, 0, :decode_len[0][0]].tolist(), [1, 0])
self.assertEqual(
decode_result.numpy()[1, 0, :decode_len[1][0]].tolist(), [0, 1, 0])
self.assertEqual(alignments.numpy()[0, 0, :decode_len[0][0]].tolist(),
[0, 4])
self.assertEqual(alignments.numpy()[1, 0, :decode_len[1][0]].tolist(),
[0, 2, 4])
np.testing.assert_almost_equal(scores.numpy(),
np.array([[-3.58212], [-3.77783]]), 5)
def test_real_ctc_decode(self):
data = np.genfromtxt("data/rnnOutput.csv", delimiter=';')[:, :-1]
inputs = np.array(
[data[t, :][np.newaxis, :] for t in range(data.shape[0])])
# Pad to max_time_steps = 8
# + 2 * [-5*np.ones(
# (1, 80), dtype=np.float32)], dtype=np.float32)
seq_lens = np.array([inputs.shape[0]], dtype=np.int32)
th_input = torch.from_numpy(inputs).type(torch.FloatTensor)
th_input = log_softmax(Variable(th_input), dim=2).data
th_seq_len = torch.IntTensor(seq_lens)
labels = ' !"#&\'()*+,-./0123456789:;?ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz_'
# greedy using beam
scorer = ctcdecode.Scorer()
decoder = ctcdecode.CTCBeamDecoder(scorer,
labels,
blank_index=labels.index('_'),
space_index=labels.index(' '),
top_paths=1,
beam_width=1)
decode_result, scores, decode_len, alignments, char_probs = decoder.decode(
th_input, th_seq_len)
txt_result = ''.join(
[labels[x] for x in decode_result[0][0][0:decode_len[0][0]]])
self.assertEqual("the fak friend of the fomly hae tC", txt_result)
# default beam decoding
decoder = ctcdecode.CTCBeamDecoder(scorer,
labels,
blank_index=labels.index('_'),
space_index=labels.index(' '),
top_paths=1,
beam_width=25)
decode_result, scores, decode_len, alignments, char_probs = decoder.decode(
th_input, th_seq_len)
txt_result = ''.join(
[labels[x] for x in decode_result[0][0][0:decode_len[0][0]]])
self.assertEqual("the fak friend of the fomcly hae tC", txt_result)
# dictionary-based decoding where non-words and words are equiprobable. Equivalent to standard beam decoding
scorer = ctcdecode.DictScorer(labels,
"data/ocr.trie",
blank_index=labels.index('_'),
space_index=labels.index(' '))
scorer.set_min_unigram_weight(0.0)
decoder = ctcdecode.CTCBeamDecoder(scorer,
labels,
blank_index=labels.index('_'),
space_index=labels.index(' '),
top_paths=1,
beam_width=25)
decode_result, scores, decode_len, alignments, char_probs = decoder.decode(
th_input, th_seq_len)
txt_result = ''.join(
[labels[x] for x in decode_result[0][0][0:decode_len[0][0]]])
self.assertEqual("the fak friend of the fomcly hae tC", txt_result)
# dictionary-based decoding - only dictionary words can be emitted
scorer = ctcdecode.DictScorer(labels,
"data/ocr.trie",
blank_index=labels.index('_'),
space_index=labels.index(' '))
decoder = ctcdecode.CTCBeamDecoder(scorer,
labels,
blank_index=labels.index('_'),
space_index=labels.index(' '),
top_paths=1,
beam_width=25)
decode_result, scores, decode_len, alignments, char_probs = decoder.decode(
th_input, th_seq_len)
txt_result = ''.join(
[labels[x] for x in decode_result[0][0][0:decode_len[0][0]]])
self.assertEqual("the fake friend of the family, fake the", txt_result)
# lm-based decoding
scorer = ctcdecode.KenLMScorer(labels,
"data/bigram.arpa",
"data/ocr.trie",
blank_index=labels.index('_'),
space_index=labels.index(' '))
scorer.set_lm_weight(2.0)
scorer.set_word_weight(0)
decoder = ctcdecode.CTCBeamDecoder(scorer,
labels,
blank_index=labels.index('_'),
space_index=labels.index(' '),
top_paths=1,
beam_width=25)
decode_result, scores, decode_len, alignments, char_probs = decoder.decode(
th_input, th_seq_len)
txt_result = ''.join(
[labels[x] for x in decode_result[0][0][0:decode_len[0][0]]])
self.assertEqual("the fake friend of the family, like the", txt_result)