File size: 5,780 Bytes
10ed3aa | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 | import copy
from contextlib import closing
import json
from pathlib import Path
import tempfile
import unittest
import numpy as np
from segment_eq255 import (check_metadata, check_scope, check_totals, decode_frames, encode_frames, label_row,
new_metadata, scope_sequences, selected, LABEL)
def fixture():
return {"dataset_id": "FreiHAND", "split": "train", "num_frames": 1,
"components": {"mano": {"frozen": True}, "segment": {
"schema_id": "segment/foreground_intensity/v1", "entity_id": "hand_right",
"value_range": [0, 255], "normalization_scale": 1 / 255, "foreground_category": "hand",
"valid_ref": "annotation/segment/mask_valid.pt", "provenance": {"kind": "format_conversion"},
"storage": {"pixel_format": "gray", "codec": "ffv1", "source_lossy": True}}}}
class TestSelection(unittest.TestCase):
def test_every_possible_value(self):
a = np.arange(256, dtype=np.uint8).reshape(16, 16)
b = selected(a)
self.assertEqual(b.dtype, np.uint8)
self.assertEqual(int(np.count_nonzero(b)), 1)
self.assertEqual(b[-1, -1], 255)
self.assertEqual(a[0, 1], 1)
def test_widened_native_input(self):
a = np.array([[0, 1, 127, 248, 254, 255]], dtype=np.uint16)
np.testing.assert_array_equal(selected(a), [[0, 0, 0, 0, 0, 255]])
def test_reject_scaling_or_float(self):
for a in (np.array([[65535]], dtype=np.uint16), np.array([[1.0]], dtype=np.float32)):
with self.assertRaises(ValueError):
selected(a)
def test_no_morphological_changes(self):
a = np.array([[255, 254, 255], [255, 0, 255], [0, 255, 0]], dtype=np.uint8)
np.testing.assert_array_equal(selected(a) == 255, a == 255)
def test_zero_is_not_mapped_to_background(self):
self.assertEqual(label_row(2, False), {"frame_index": 2, "labels": []})
self.assertEqual(label_row(3, True)["labels"], [LABEL])
self.assertEqual(LABEL["entity_id"], "hand_right")
self.assertIsNone(LABEL["source_instance_id"])
def test_metadata_only_segment(self):
a = fixture()
old = copy.deepcopy(a)
b = new_metadata(a)
self.assertEqual(a, old)
check_metadata(a, b)
self.assertEqual(b["components"]["segment"]["schema_id"], "segment/uint16_labels/v1")
def test_other_annotation_mutation_rejected(self):
a = fixture()
b = new_metadata(a)
b["components"]["mano"]["frozen"] = False
with self.assertRaises(ValueError):
check_metadata(a, b)
def test_validity_reference_mutation_rejected(self):
a = fixture()
b = new_metadata(a)
b["components"]["segment"]["valid_ref"] = "changed.pt"
with self.assertRaises(ValueError):
check_metadata(a, b)
def test_eval_conversion_rejected(self):
a = fixture()
a["split"] = "test"
a["components"]["segment"]["schema_id"] = "segment/uint16_labels/v1"
with self.assertRaises(ValueError):
new_metadata(a)
def test_decoder_and_encoder_declarations_checked(self):
a = fixture()
for key, value in (("decoder", "scale gray8 to gray16le"),
("encoder_parameters", {"level": 3, "threads": 1, "input_pixel_format": "gray16le"})):
b = new_metadata(a)
b["components"]["segment"]["storage"][key] = value
with self.assertRaises(ValueError):
check_metadata(a, b)
def test_exact_pilot_and_full_scope(self):
self.assertEqual(scope_sequences("pilot"), ["evaluation_00000000", "training_00000000", "training_00126976"])
self.assertEqual(len(scope_sequences("all_33_shards")), 33)
with self.assertRaises(ValueError):
scope_sequences("invented")
def test_partial_full_claim_rejected(self):
with self.assertRaises(ValueError):
check_scope(Path("/not-used"), {"scope": "all_33_shards", "sequences": ["training_00000000"]})
def test_duplicate_or_missing_eval_rejected(self):
for seqs in (["training_00000000", "training_00126976"], scope_sequences("pilot") + ["training_00000000"]):
with self.assertRaises(ValueError):
check_scope(Path("/not-used"), {"scope": "pilot", "sequences": seqs})
def test_summary_recomputed(self):
views = [{"frames": 5, "pixels_checked": 500, "empty_after_selection_frames": [2]}]
good = {"training_frames": 5, "pixels_checked": 500, "empty_frames": 1}
check_totals(good, views)
for key in good:
bad = good | {key: good[key] + 1}
with self.assertRaises(ValueError):
check_totals(bad, views)
def test_lossless_gray8_video_roundtrip(self):
a = np.arange(256, dtype=np.uint8).reshape(16, 16)
frames = [selected(a), np.zeros_like(a), np.full_like(a, 255)]
with tempfile.TemporaryDirectory() as d:
path = Path(d) / "mask.mkv"
encode_frames(path, iter(frames), 16, 16)
with closing(decode_frames(path, "gray", 16, 16, 3)) as stream:
actual = list(stream)
for x, y in zip(frames, actual, strict=True):
self.assertEqual(y.dtype, np.uint16)
np.testing.assert_array_equal(x, y)
with self.assertRaises(ValueError):
with closing(decode_frames(path, "gray", 16, 16, 2)) as stream:
list(stream)
with self.assertRaises(ValueError):
with closing(decode_frames(path, "gray", 16, 16, 4)) as stream:
list(stream)
if __name__ == "__main__":
unittest.main()
|