Handle verified Ego4D terminal bounding-box records
This commit is contained in:
Xin Wang
163
test_ego4d_box_frames.py
Normal file
163
test_ego4d_box_frames.py
Normal file
@@ -0,0 +1,163 @@
|
||||
"""CPU regressions for a verified Ego4D exclusive-end bbox record."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import contextlib
|
||||
import copy
|
||||
import io
|
||||
import json
|
||||
from pathlib import Path
|
||||
import sys
|
||||
import tempfile
|
||||
from types import SimpleNamespace
|
||||
import unittest
|
||||
from unittest.mock import patch
|
||||
|
||||
import numpy as np
|
||||
from PIL import Image
|
||||
import torch
|
||||
|
||||
import extract_turn_local_features as extract
|
||||
from local_turn_features import load_local_features
|
||||
from test_local_turn_features import fixture
|
||||
|
||||
|
||||
UID = "31488ff6-05ef-4362-9786-35981b06f307"
|
||||
|
||||
|
||||
def terminal_clip():
|
||||
return {
|
||||
"clip_uid": UID, "clip_start_frame": 0, "clip_end_frame": 8999,
|
||||
"clip_start_sec": 0, "clip_end_sec": 299.9666666666666,
|
||||
"persons": [{"person_id": "p1", "tracking_paths": [{"track": [
|
||||
{"frame": frame, "clip_frame": None, "x": 0, "y": 0, "width": 2, "height": 2}
|
||||
for frame in (0, 8985, 8998, 8999)
|
||||
]}]}],
|
||||
}
|
||||
|
||||
|
||||
def terminal_annotation():
|
||||
clip = terminal_clip()
|
||||
return {"frames": extract.ego4d_clip_boxes(clip),
|
||||
"bounds": {key: clip[key] for key in (
|
||||
"clip_start_frame", "clip_end_frame", "clip_start_sec", "clip_end_sec",
|
||||
)}}
|
||||
|
||||
|
||||
class Ego4DBoxFrameTests(unittest.TestCase):
|
||||
def setUp(self):
|
||||
torch.set_num_threads(1)
|
||||
self.source = SimpleNamespace(num_frames=8999, source_fps=30.)
|
||||
|
||||
def test_confirmed_terminal_only_is_removed_without_mutating_input(self):
|
||||
annotation = terminal_annotation()
|
||||
original = copy.deepcopy(annotation)
|
||||
with contextlib.redirect_stdout(io.StringIO()) as log:
|
||||
frames, audit = extract.validate_ego4d_box_frames(UID, annotation, self.source)
|
||||
self.assertEqual(frames, {frame: original["frames"][frame] for frame in (0, 8985, 8998)})
|
||||
self.assertEqual(annotation, original)
|
||||
self.assertEqual(audit["ignored_frame_indices"], [8999])
|
||||
self.assertEqual(audit["ignored_box_count"], 1)
|
||||
self.assertEqual(audit["annotation_frame_range"], [0, 8999])
|
||||
self.assertIn("No frame shift or feature/target row removal", log.getvalue())
|
||||
|
||||
def test_valid_and_empty_mappings_are_unchanged_even_without_bounds(self):
|
||||
for frames in ({}, {0: [], 8998: [{"id": "valid"}]}):
|
||||
with self.subTest(frames=frames):
|
||||
result, audit = extract.validate_ego4d_box_frames(
|
||||
UID, {"frames": frames, "bounds": {}}, self.source)
|
||||
self.assertIs(result, frames)
|
||||
self.assertEqual(audit["ignored_frame_indices"], [])
|
||||
|
||||
def test_larger_overrun_negative_and_truncated_source_still_fail(self):
|
||||
for extra in (-1, 9000, 12000):
|
||||
annotation = terminal_annotation()
|
||||
annotation["frames"][extra] = []
|
||||
with self.subTest(frame=extra), self.assertRaisesRegex(ValueError, "source_frames=8999"):
|
||||
extract.validate_ego4d_box_frames(UID, annotation, self.source)
|
||||
with self.assertRaisesRegex(ValueError, "source_frames=8998"):
|
||||
extract.validate_ego4d_box_frames(
|
||||
UID, terminal_annotation(), SimpleNamespace(num_frames=8998, source_fps=30.))
|
||||
|
||||
def test_terminal_requires_zero_origin_matching_declared_end_and_duration(self):
|
||||
for key, value in (("clip_start_frame", 1), ("clip_end_frame", 9000),
|
||||
("clip_start_sec", .1), ("clip_end_sec", 300.),
|
||||
("clip_end_sec", float("nan"))):
|
||||
annotation = terminal_annotation()
|
||||
annotation["bounds"][key] = value
|
||||
with self.subTest(key=key, value=value), self.assertRaisesRegex(ValueError, "declared_bounds"):
|
||||
extract.validate_ego4d_box_frames(UID, annotation, self.source)
|
||||
for key in terminal_annotation()["bounds"]:
|
||||
annotation = terminal_annotation()
|
||||
del annotation["bounds"][key]
|
||||
with self.subTest(missing=key), self.assertRaises(ValueError):
|
||||
extract.validate_ego4d_box_frames(UID, annotation, self.source)
|
||||
for fps in (29.97, 15., 0., float("nan")):
|
||||
with self.subTest(fps=fps), self.assertRaises(ValueError):
|
||||
extract.validate_ego4d_box_frames(
|
||||
UID, terminal_annotation(), SimpleNamespace(num_frames=8999, source_fps=fps))
|
||||
|
||||
def test_full_extraction_keeps_1499_rows_saves_audit_and_reuses(self):
|
||||
image = Image.new("RGB", (2, 2))
|
||||
source = SimpleNamespace(num_frames=8999, source_fps=30., crop=None, dense_start=0.,
|
||||
path=Path(f"{UID}.mp4"), width=2, height=2)
|
||||
media = SimpleNamespace(
|
||||
clips=[source], metadata=lambda: {},
|
||||
_source_frame_for_time=lambda t: (0, int(t * 30 + 1e-8), int(t * 30 + 1e-8) / 30),
|
||||
frame_at_or_before=lambda t: image,
|
||||
audio_slice=lambda start, end: np.array([start, end]),
|
||||
)
|
||||
fake_module = SimpleNamespace(CausalSourceAV=lambda *a, **kw: media,
|
||||
SourceClipSpec=lambda path: path, _easycom_directory_specs=None)
|
||||
encoder = SimpleNamespace(
|
||||
visual_dim=6, audio_dim=5, transformers_version="4.57.6", model_config_sha256="test",
|
||||
images=lambda crops: torch.ones(len(crops), 6),
|
||||
audio=lambda waveform: torch.full((5,), float(waveform[-1])),
|
||||
)
|
||||
with tempfile.TemporaryDirectory() as temporary:
|
||||
root = Path(temporary)
|
||||
for folder in ("global", "targets", "v1/clips", "v1/annotations"):
|
||||
(root / folder).mkdir(parents=True)
|
||||
reference, _, targets = fixture(n=1499, name=UID, dataset="ego4d")
|
||||
torch.save(reference, root / "global" / f"{UID}.pt")
|
||||
np.save(root / "targets" / f"{UID}.npy", targets)
|
||||
(root / "v1/clips" / f"{UID}.mp4").write_bytes(b"fake-video")
|
||||
for split, clips in (("train", [terminal_clip()]), ("val", [])):
|
||||
(root / f"v1/annotations/av_{split}.json").write_text(json.dumps({"videos": [{"clips": clips}]}))
|
||||
loaded = extract.load_ego4d_boxes([root / "v1/annotations/av_train.json"], [UID])
|
||||
self.assertEqual(loaded[UID], terminal_annotation())
|
||||
argv = ["--dataset", "ego4d", "--data-root", str(root), "--feature-root", str(root / "global"),
|
||||
"--target-root", str(root / "targets"), "--output-root", str(root / "local"),
|
||||
"--samples", UID, "--no-progress"]
|
||||
with patch.dict(sys.modules, {"transformers": SimpleNamespace(__version__="4.57.6"),
|
||||
"streaming_turn_model_directly": fake_module}), \
|
||||
patch.object(extract, "FrozenQwenLocalEncoder", return_value=encoder) as model, \
|
||||
contextlib.redirect_stdout(io.StringIO()):
|
||||
extract.main(argv)
|
||||
model.assert_called_once()
|
||||
destination = root / "local" / f"{UID}.pt"
|
||||
result = torch.load(destination, weights_only=False)
|
||||
self.assertEqual(result["visual"].shape, (1499, 6))
|
||||
self.assertEqual(result["audio"].shape, (1499, 5))
|
||||
self.assertTrue(torch.equal(result["times"], reference["times"]))
|
||||
self.assertTrue(torch.equal(result["label_indices"], reference["label_indices"]))
|
||||
self.assertAlmostEqual(result["times"][-1].item(), 299.8)
|
||||
self.assertFalse(result["partial"])
|
||||
self.assertEqual(result["box_frame_audit"]["ignored_frame_indices"], [8999])
|
||||
self.assertEqual(result["persons"]["frame_times"][-1].item(), 299.5)
|
||||
self.assertEqual(result["persons"]["records"][-1]["source_frame_index"], 8985)
|
||||
load_local_features(destination, reference, sample_id=UID, dataset="ego4d", modalities="av")
|
||||
# Previously completed valid sidecars have no audit. Their
|
||||
# vectors/config remain reusable: only unsampled boxes changed.
|
||||
del result["box_frame_audit"]
|
||||
torch.save(result, destination)
|
||||
model.reset_mock()
|
||||
extract.main(argv + ["--reuse-existing"])
|
||||
model.assert_not_called()
|
||||
extract.main(argv + ["--preview-only", "--preview-count", "1"])
|
||||
audit = json.loads((root / "local" / f"{UID}_preview/box_frame_audit.json").read_text())
|
||||
self.assertEqual(audit["ignored_frame_indices"], [8999])
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
Reference in New Issue
Block a user