File size: 20,666 Bytes
699f3cd | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 | from __future__ import annotations
import hashlib
import json
import logging
from dataclasses import dataclass, field
from pathlib import Path
from typing import Any, ClassVar
from src.core.vision import is_cv2_available
from src.task3.routing_policy import assign_detector
if is_cv2_available(): # pragma: no branch - ortama bagli
from src.core.vision import cv2, np
else: # pragma: no cover - cv2 yoksa
cv2 = None
np = None
LOGGER = logging.getLogger(__name__)
@dataclass(slots=True)
class ReferenceCache:
"""Referans nesne metadata ve descriptor cache iskeleti."""
items: dict[str, dict[str, Any]] = field(default_factory=dict)
auto_routing_summary: dict[str, dict[str, Any]] = field(default_factory=dict)
overrides_applied: list[dict[str, Any]] = field(default_factory=list)
per_reference_suppression: bool = False
IMAGE_EXTENSIONS: ClassVar[tuple[str, ...]] = (".jpg", ".jpeg", ".png", ".bmp", ".webp", ".pgm")
VALID_DETECTORS: ClassVar[set[str]] = {"yoloe", "orb", "both"}
VALID_MODALITIES: ClassVar[set[str]] = {"rgb", "thermal", "unknown"}
VALID_OVERRIDE_FIELDS: ClassVar[set[str]] = {"detector", "detector_modalities", "modality", "rationale"}
def put(self, reference_id: str, payload: dict[str, Any]) -> None:
self.items[reference_id] = payload
def get(self, reference_id: str) -> dict[str, Any] | None:
return self.items.get(reference_id)
def list_ids(self) -> list[str]:
return sorted(self.items.keys())
def get_auto_routing_summary(self) -> dict[str, dict[str, Any]]:
return {reference_id: dict(payload) for reference_id, payload in self.auto_routing_summary.items()}
def get_routing_diagnostics(self) -> dict[str, dict[str, Any]]:
diagnostics: dict[str, dict[str, Any]] = {}
for reference_id in self.list_ids():
item = self.get(reference_id) or {}
metadata = dict(item.get("reference_metadata") or {})
diagnostics[reference_id] = {
"modality": str(item.get("reference_modality") or metadata.get("modality") or "unknown"),
"detector": str(item.get("detector") or metadata.get("detector") or "yoloe"),
"detector_modalities": list(item.get("detector_modalities") or metadata.get("detector_modalities") or []),
"confidence": str(metadata.get("routing_confidence") or "low"),
"rationale": str(metadata.get("routing_rationale") or ""),
"signals": dict(metadata.get("routing_signals") or {}),
"override": dict(metadata.get("override") or {}) if metadata.get("override") else None,
}
return diagnostics
def get_overrides_applied(self) -> list[dict[str, Any]]:
return [dict(item) for item in self.overrides_applied]
def get_candidate_suppression_mode(self) -> str:
return "per_reference_top_1" if self.per_reference_suppression else "global_top_1"
def set_learned_embedding(self, reference_id: str, backbone_name: str, embedding: Any) -> None:
if reference_id not in self.items:
return
self.items[reference_id].setdefault("learned_embeddings", {})
self.items[reference_id]["learned_embeddings"][backbone_name] = embedding
def get_learned_embedding(self, reference_id: str, backbone_name: str) -> Any | None:
item = self.get(reference_id)
if not item:
return None
return item.get("learned_embeddings", {}).get(backbone_name)
def get_detector(self, reference_id: str, *, default: str = "yoloe") -> str:
item = self.get(reference_id) or {}
detector = str(item.get("detector") or default)
return detector
def get_detector_modalities(self, reference_id: str) -> tuple[str, ...] | None:
item = self.get(reference_id) or {}
raw_modalities = item.get("detector_modalities")
if not raw_modalities:
return None
return tuple(str(modality) for modality in raw_modalities)
def allows_detector_for_modality(self, reference_id: str, modality: str | None) -> bool:
allowed_modalities = self.get_detector_modalities(reference_id)
if not allowed_modalities or not modality or str(modality) == "unknown":
return True
return str(modality) in allowed_modalities
def filter_reference_ids_by_detector_modality(
self,
reference_ids: list[str],
*,
modality: str | None,
) -> list[str]:
return [reference_id for reference_id in reference_ids if self.allows_detector_for_modality(reference_id, modality)]
def split_reference_ids_by_detector(
self,
reference_ids: list[str],
*,
default_detector: str = "yoloe",
) -> tuple[list[str], list[str]]:
yoloe_ids: list[str] = []
orb_ids: list[str] = []
for reference_id in reference_ids:
detector = self.get_detector(reference_id, default=default_detector)
if detector == "both":
yoloe_ids.append(reference_id)
orb_ids.append(reference_id)
elif detector == "orb":
orb_ids.append(reference_id)
else:
yoloe_ids.append(reference_id)
return yoloe_ids, orb_ids
def preload_from_directory(self, path: str | Path, *, orb_features: int = 256) -> int:
directory = Path(path)
if not directory.exists() or not directory.is_dir():
return 0
self.items.clear()
self.auto_routing_summary.clear()
self.overrides_applied.clear()
self.per_reference_suppression = False
spec_payload = self._load_reference_spec(directory)
reference_metadata = spec_payload["references"]
overrides = spec_payload["overrides"]
auto_routing_enabled = bool(spec_payload.get("auto_routing_enabled"))
self.per_reference_suppression = bool(spec_payload.get("per_reference_suppression", False))
file_to_reference_id = {
str(metadata.get("file", "")).lower(): reference_id
for reference_id, metadata in reference_metadata.items()
if str(metadata.get("file", "")).strip()
}
loaded_count = 0
orb = cv2.ORB_create(nfeatures=orb_features) if is_cv2_available() else None
for candidate in sorted(directory.iterdir()):
if not candidate.is_file():
continue
if candidate.suffix.lower() not in self.IMAGE_EXTENSIONS:
continue
data = candidate.read_bytes()
reference_id = file_to_reference_id.get(candidate.name.lower(), candidate.stem)
metadata = dict(reference_metadata.get(reference_id, {}))
try:
if auto_routing_enabled:
assignment = assign_detector(candidate)
effective_detector = assignment.detector
effective_modalities = list(assignment.detector_modalities)
effective_modality = str(assignment.signals.get("modality", metadata.get("modality") or "unknown"))
effective_confidence = str(assignment.confidence)
effective_rationale = str(assignment.rationale)
routing_signals = dict(assignment.signals)
else:
effective_detector = "yoloe"
effective_modalities = None
effective_modality = str(metadata.get("modality") or "unknown")
effective_confidence = "low"
effective_rationale = "specless preload; auto-routing skipped and detector defaulted to yoloe"
routing_signals = {
"modality": effective_modality,
"modality_method": "none",
"modality_confidence": "low",
"modality_reason": "manifest/spec unavailable",
"modality_exif_signals": [],
"modality_pixel_signals": {},
}
except Exception as exc:
effective_detector = "yoloe"
effective_modalities = None
effective_modality = "unknown"
effective_confidence = "low"
effective_rationale = f"auto-routing unavailable; defaulted to yoloe ({exc})"
routing_signals = {
"modality": "unknown",
"modality_method": "none",
"modality_confidence": "low",
"modality_reason": str(exc),
"modality_exif_signals": [],
"modality_pixel_signals": {},
}
LOGGER.warning("reference %s auto-routing failed: %s", reference_id, exc)
manual_override = overrides.get(reference_id)
override_summary: dict[str, Any] | None = None
if manual_override is not None:
auto_assignment_summary = {
"detector": effective_detector,
"detector_modalities": list(effective_modalities) if effective_modalities is not None else None,
"modality": effective_modality,
}
effective_detector = str(manual_override["detector"])
if "detector_modalities" in manual_override:
effective_modalities = list(manual_override["detector_modalities"])
if manual_override.get("modality") is not None:
effective_modality = str(manual_override["modality"])
effective_confidence = "medium" if effective_confidence == "high" else effective_confidence
override_rationale = str(manual_override.get("rationale") or "manual override")
effective_rationale = f"{override_rationale}"
override_summary = {
"reference_id": reference_id,
"auto_detected": auto_assignment_summary,
"override": {
"detector": effective_detector,
"detector_modalities": list(effective_modalities) if effective_modalities is not None else None,
"modality": effective_modality,
"rationale": override_rationale,
},
}
self.overrides_applied.append(override_summary)
LOGGER.warning(
"reference %s has manual override; auto-detected was %s, overridden to %s (%s)",
reference_id,
auto_assignment_summary,
override_summary["override"],
override_rationale,
)
auto_summary_payload = {
"modality": effective_modality,
"detector": effective_detector,
"detector_modalities": list(effective_modalities) if effective_modalities is not None else None,
"confidence": effective_confidence,
"rationale": effective_rationale,
}
self.auto_routing_summary[reference_id] = auto_summary_payload
gray = None
bgr = None
descriptors = None
keypoints = None
width = 0
height = 0
descriptor_mode = "metadata_only"
if orb is not None:
bgr = cv2.imdecode(np.frombuffer(data, dtype=np.uint8), cv2.IMREAD_COLOR)
if bgr is not None:
gray = cv2.cvtColor(bgr, cv2.COLOR_BGR2GRAY)
height, width = gray.shape[:2]
keypoints, descriptors = orb.detectAndCompute(gray, None)
descriptor_mode = "orb"
self.put(
reference_id,
{
"reference_id": reference_id,
"path": str(candidate),
"byte_size": len(data),
"sha1": hashlib.sha1(data).hexdigest(),
"loaded_from": "directory_preload",
"bgr": bgr,
"gray": gray,
"width": width,
"height": height,
"keypoints": keypoints,
"descriptors": descriptors,
"descriptor_mode": descriptor_mode,
"keypoint_count": len(keypoints or []),
"learned_embeddings": {},
"detector": effective_detector,
"detector_modalities": list(effective_modalities) if effective_modalities is not None else None,
"reference_modality": effective_modality,
"reference_dimensions": metadata.get("dimensions"),
"reference_metadata": {
**metadata,
"modality": effective_modality,
"detector": effective_detector,
"detector_modalities": list(effective_modalities) if effective_modalities is not None else None,
"routing_confidence": effective_confidence,
"routing_rationale": effective_rationale,
"routing_signals": routing_signals,
"override": override_summary["override"] if override_summary else None,
},
},
)
loaded_count += 1
return loaded_count
def _load_reference_spec(self, directory: Path) -> dict[str, dict[str, Any]]:
manifest_path = directory / "manifest.json"
if not manifest_path.exists():
return {"references": {}, "overrides": {}, "auto_routing_enabled": False, "per_reference_suppression": False}
manifest_payload = json.loads(manifest_path.read_text(encoding="utf-8"))
spec_path_raw = str(manifest_payload.get("spec_path") or "").strip()
if not spec_path_raw:
return {"references": {}, "overrides": {}, "auto_routing_enabled": False, "per_reference_suppression": False}
spec_path = Path(spec_path_raw)
if not spec_path.is_absolute():
candidate = directory / spec_path_raw
spec_path = candidate if candidate.exists() else Path(spec_path_raw)
if not spec_path.exists():
return {"references": {}, "overrides": {}, "auto_routing_enabled": False, "per_reference_suppression": False}
spec_payload = json.loads(spec_path.read_text(encoding="utf-8"))
references = spec_payload.get("references", {})
if not isinstance(references, dict):
raise ValueError(f"Task3 reference spec malformed: {spec_path}")
per_reference_suppression = spec_payload.get("per_reference_suppression", False)
if not isinstance(per_reference_suppression, bool):
raise ValueError(f"Task3 reference spec per_reference_suppression must be boolean: {spec_path}")
overrides_payload = spec_payload.get("overrides", {})
if overrides_payload is None:
overrides_payload = {}
if not isinstance(overrides_payload, dict):
raise ValueError(f"Task3 reference spec overrides malformed: {spec_path}")
metadata_by_reference: dict[str, dict[str, Any]] = {}
nested_overrides: dict[str, dict[str, Any]] = {}
for reference_id, payload in references.items():
if not isinstance(payload, dict):
raise ValueError(f"Task3 reference spec entry malformed for {reference_id}: {spec_path}")
reference_id_text = str(reference_id)
routing_override = payload.get("routing_override")
if routing_override is not None:
nested_overrides[reference_id_text] = self._normalize_override_payload(
routing_override,
reference_id_text,
spec_path=spec_path,
)
metadata_by_reference[reference_id_text] = {
"modality": payload.get("modality"),
"dimensions": payload.get("dimensions"),
"source_exif": payload.get("source_exif"),
"file": payload.get("file"),
}
normalized_overrides: dict[str, dict[str, Any]] = {}
for reference_id, payload in overrides_payload.items():
reference_id_text = str(reference_id)
normalized_overrides[reference_id_text] = self._normalize_override_payload(
payload,
reference_id_text,
spec_path=spec_path,
)
for reference_id, payload in nested_overrides.items():
if reference_id in normalized_overrides:
raise ValueError(
f"Task3 reference spec override duplicated for {reference_id}: "
f"use either top-level overrides or references.{reference_id}.routing_override"
)
normalized_overrides[reference_id] = payload
return {
"references": metadata_by_reference,
"overrides": normalized_overrides,
"auto_routing_enabled": True,
"per_reference_suppression": per_reference_suppression,
}
def _normalize_override_payload(
self,
payload: Any,
reference_id: str,
*,
spec_path: Path,
) -> dict[str, Any]:
if not isinstance(payload, dict):
raise ValueError(f"Task3 reference spec override malformed for {reference_id}: {spec_path}")
unexpected_keys = sorted(set(payload.keys()) - self.VALID_OVERRIDE_FIELDS)
if unexpected_keys:
raise ValueError(
f"Task3 reference spec override contains unsupported fields for {reference_id}: {unexpected_keys}"
)
if "detector" not in payload:
raise ValueError(f"Task3 reference spec override missing detector for {reference_id}: {spec_path}")
detector = self._validate_detector(str(payload["detector"]), reference_id)
modality = payload.get("modality")
if modality is not None and str(modality) not in self.VALID_MODALITIES:
raise ValueError(
f"Task3 reference spec override modality must be within {sorted(self.VALID_MODALITIES)}, "
f"got {modality!r} for {reference_id}"
)
detector_modalities = self._normalize_detector_modalities(payload.get("detector_modalities"), reference_id)
if detector_modalities is None and modality is not None and str(modality) in {"rgb", "thermal"}:
detector_modalities = [str(modality)]
return {
"detector": detector,
"detector_modalities": detector_modalities,
"modality": str(modality) if modality is not None else None,
"rationale": payload.get("rationale"),
}
def _validate_detector(self, detector: str, reference_id: str) -> str:
if detector not in self.VALID_DETECTORS:
raise ValueError(
f"Task3 reference spec detector must be one of {sorted(self.VALID_DETECTORS)}, "
f"got {detector!r} for {reference_id}"
)
return detector
def _normalize_detector_modalities(self, detector_modalities: Any, reference_id: str) -> list[str] | None:
if detector_modalities is None:
return None
if not isinstance(detector_modalities, list) or not detector_modalities:
raise ValueError(f"Task3 reference spec detector_modalities must be a non-empty list for {reference_id}")
normalized_modalities: list[str] = []
for modality in detector_modalities:
modality_text = str(modality)
if modality_text not in self.VALID_MODALITIES:
raise ValueError(
f"Task3 reference spec detector_modalities must be within {sorted(self.VALID_MODALITIES)}, "
f"got {modality_text!r} for {reference_id}"
)
if modality_text not in normalized_modalities:
normalized_modalities.append(modality_text)
return normalized_modalities
|