dsh/dual_camera_system/core/capture_uploader.py

"""检测到人后的保存 + 上传."""
import logging
import os
import threading
import time
import uuid
from typing import Any, Callable, Dict, List, Optional
import cv2
import numpy as np

from config.device import DEVICE_CONFIG

logger = logging.getLogger(__name__)


class CaptureUploader:
    def __init__(
        self,
        group_id: str,
        save_dir: str = "data/captures",
        upload_callback: Optional[Callable[[Dict], None]] = None,
        dedup_seconds: float = 5.0,
    ):
        self.group_id = group_id
        self.save_dir = os.path.join(save_dir, group_id)
        os.makedirs(self.save_dir, exist_ok=True)
        self.upload_callback = upload_callback
        self.dedup_seconds = dedup_seconds
        self._last_uploads: List[Dict[str, Any]] = []
        self._lock = threading.Lock()
        self._counter = 0

    def _should_upload(self, camera_type: str, bbox: List[float]) -> bool:
        cx = (bbox[0] + bbox[2]) / 2
        cy = (bbox[1] + bbox[3]) / 2
        for u in self._last_uploads:
            if u["camera_type"] != camera_type:
                continue
            dx = abs(u["cx"] - cx)
            dy = abs(u["cy"] - cy)
            if dx < 50 and dy < 50:
                return False
        return True

    def _validate_inputs(
        self,
        camera_type: str,
        frame: np.ndarray,
        detections: List[Dict],
    ) -> None:
        if camera_type not in ("panorama", "ptz"):
            raise ValueError("camera_type must be 'panorama' or 'ptz'")

        if not isinstance(frame, np.ndarray):
            raise ValueError("frame must be a numpy ndarray")
        if frame.ndim != 3 or frame.shape[2] != 3:
            raise ValueError("frame must have shape (H, W, 3)")
        if frame.dtype != np.uint8:
            raise ValueError("frame must have dtype uint8")

        for i, det in enumerate(detections):
            if not isinstance(det, dict):
                raise ValueError(f"detection {i} must be a dict")
            if "bbox" not in det:
                raise ValueError(f"detection {i} missing bbox")
            bbox = det["bbox"]
            if not isinstance(bbox, (list, tuple)) or len(bbox) != 4:
                raise ValueError(f"detection {i} bbox must be a list/tuple of 4 numbers")
            try:
                [float(v) for v in bbox]
            except (TypeError, ValueError):
                raise ValueError(f"detection {i} bbox must contain numbers")
            if "confidence" not in det:
                raise ValueError(f"detection {i} missing confidence")
            try:
                float(det["confidence"])
            except (TypeError, ValueError):
                raise ValueError(f"detection {i} confidence must be a number")

    def handle_detection(
        self,
        camera_type: str,  # 'panorama' or 'ptz'
        frame: np.ndarray,
        detections: List[Dict],
        ptz_position: Optional[Dict] = None,
    ) -> List[Dict]:
        self._validate_inputs(camera_type, frame, detections)

        if not detections:
            return []

        with self._lock:
            now = time.monotonic()
            self._last_uploads = [
                u for u in self._last_uploads
                if now - u["time"] < self.dedup_seconds
            ]

            upload_decisions = [
                (det, self._should_upload(camera_type, det["bbox"]))
                for det in detections
            ]
            to_upload = [det for det, should in upload_decisions if should]
            if not to_upload:
                logger.debug("All detections deduplicated; skipping file writes for %s", camera_type)
                return []

            self._counter += 1
            counter = self._counter
            ts = int(time.time() * 1000)
            original_path = os.path.join(
                self.save_dir, f"{camera_type}_{ts}_{counter}_original.jpg"
            )
            marked_path = os.path.join(
                self.save_dir, f"{camera_type}_{ts}_{counter}_marked.jpg"
            )

            # Reserve dedup slots and generate paths under lock.
            for det in to_upload:
                self._last_uploads.append({
                    "camera_type": camera_type,
                    "cx": (det["bbox"][0] + det["bbox"][2]) / 2,
                    "cy": (det["bbox"][1] + det["bbox"][3]) / 2,
                    "time": time.monotonic(),
                })

        # File I/O and user callback run outside the lock.
        logger.info("Saving original image to %s", original_path)
        if not cv2.imwrite(original_path, frame):
            raise RuntimeError(f"Failed to write {original_path}")

        marked = frame.copy()
        for det in to_upload:
            x1, y1, x2, y2 = map(int, det["bbox"])
            cv2.rectangle(marked, (x1, y1), (x2, y2), (0, 255, 0), 2)
            cv2.putText(marked, f"{det['confidence']:.2f}", (x1, y1 - 5),
                        cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 0), 2)
        logger.info("Saving marked image to %s", marked_path)
        if not cv2.imwrite(marked_path, marked):
            raise RuntimeError(f"Failed to write {marked_path}")

        results: List[Dict] = []
        for det in to_upload:
            x1, y1, x2, y2 = map(int, det["bbox"])
            payload = {
                "group_id": self.group_id,
                "camera_type": camera_type,
                "timestamp": ts,
                "original": original_path,
                "marked": marked_path,
                "bbox": [x1, y1, x2, y2],
                "confidence": det["confidence"],
                "ptz_position": ptz_position,
            }
            results.append(payload)

        if self.upload_callback and results:
            batch_info = {
                "batch_id": str(uuid.uuid4()),
                "device_id": DEVICE_CONFIG.get("device_id", "unknown"),
                "project_id": DEVICE_CONFIG.get("project_id", ""),
                "timestamp": ts,
                "image_paths": [original_path, marked_path],
                "detections": [
                    {"bbox": det["bbox"], "confidence": det["confidence"], "camera_type": camera_type}
                    for det in to_upload
                ],
                "ptz_position": ptz_position,
            }
            logger.info("Uploading batch info for %s", camera_type)
            try:
                self.upload_callback(batch_info)
            except Exception as exc:  # noqa: BLE001
                logger.warning("Upload callback failed: %s", exc)

        return results