采集循环只认新帧，选区统计移到数值卡，状态栏改为会过期的通知行 - What: 1) 选区统计不再挤在曲线下方（中文长句会顶出面板右边界）——框选后三个数值卡的大号数字变成该窗口的峰值、下方小字为该窗口的均值·σ（chart.selection_detail），峰值语义放在卡片 tooltip；曲线下方只保留「选区 4.00s / 121 点」（完整一行留在它的 tooltip 里）。「全程总览」不算选区（那是默认适配），此时卡片回到"游标处样本"；切换单位/换数据源/返回最新都会正确切回。2) 上方的 63.0 FPS 是采集循环自己的轮询速率，根因是同一帧被反复当成新样本：CaptureSource.read_latest 返回后台线程的最新帧，没有新帧时把同一帧再给一次，循环未做判断，于是 30 fps 的设备录出约 63 行/秒（frame_id 还各不相同）。新增 _capture_sequence()：读 capture_source.status.last_seq，计数未变即"还没有新帧"，睡 2 ms 重试、不进处理链、不发 packet；只有带后台读取线程的实时相机有该标记，视频文件是同步解码、每次都取下一帧，因此忽略（行为不变）。packet.fps 改为速率 (窗口帧数-1)/(首末到达差)，窗口 1 s——按"1 秒内到达几帧"数数会系统性偏大。容器速率仍取传感器配置值，与此层无关。3) 底部状态栏只留一次性通知：删除右下角常驻「配置 sensor.json」（顶栏摘要按钮已写明），删除与顶栏重复的 status.run_started（运行中（0））与 status.connecting（正在连接…）；QStatusBar 换成 _NoticeBar——Qt 的 showMessage(text) 默认 msecs=0 表示永不消失，这正是那句「运行中（0）」一直挂到底部的原因，子类统一补 6 秒超时；status.units_raw_fallback 之前每帧都发（等于永久占住通知行），改为每次运行只提示一次。附带修复：英文下表头模式名被截断（长句移入 tooltip，标签只留模式名 chart.mode_*_tip）；数据源切换后曲线下方的选区文案会残留，改为在 _on_source_changed 里清零。 - Why: 用户评审 3 点——选区统计显示不全、顶部 fps 应受硬件限制最大 30、右下角配置文件名与左下角「运行中（0）」和顶栏信息重复。第 2 点查出根因是重复采样（数据缺陷，不只是显示问题），修掉后设备速率与记录行数才真正一致。 - Verification: ui_check.py 546 项 0 失败（新增：选区统计在卡片、卡片小字不裁切、状态栏无常驻标签且消息过期、顶栏 FPS 即 packet 交付速率、表头模式名中英不裁切）；test_ui_smoke.py 71 项 0 失败（新增选区读数与通知行过期两例）；drive_shared.py 148 项 0 失败（新增 shared/tests/test_worker_frame_rate.py：同一帧返回 3 次时 packet 数≈交付帧数，去掉判据为 11317:39，且 fps 与交付速率一致）；locale_studio.py 267 键齐全、zh==en、prefs 对齐；reproduce.py 退出 0；ui_snapshot.py 新增 --select A,B，六张复核图（选区中英、实时、录制中、回放、标定 1280x720）。 - Not in this change: 录制文件格式与 CSV 列序不变；去重只影响"哪些帧算一帧"，视频文件源行为不变；曲线下方不再显示三通道峰/均/σ（数值在卡片上，完整一行在 tooltip 里）。

2026-09-13 21:19:23 +08:00
parent 824544fc7a
commit e6a1e77a5b
13 changed files with 783 additions and 66 deletions
@@ -36,6 +36,11 @@ STOP_JOIN_TIMEOUT_S = 5.0
 #: How long a new session waits (in its own thread, never on the GUI) for a
 #: previous session that is still winding down before opening the device.
 HANDOVER_TIMEOUT_S = 60.0
+#: Window over which the delivered frame rate is measured.
+FPS_WINDOW_S = 1.0
+#: How long to wait before asking again when the capture source has not produced
+#: a new frame yet. Long enough not to spin, short enough not to add latency.
+CAPTURE_REPEAT_POLL_S = 0.002

 # Re-exported so ``from ...sdk_worker import SdkPacket`` keeps working; the
 # definitions live in ``models`` so neither worker depends on the other.
@@ -72,6 +77,29 @@ def _sensor_force_calibrated(sensor, *, config_ref: str | Path | None = None) ->
        return False


+def _capture_sequence(sensor) -> Optional[int]:
+    """The capture source's delivered-frame counter, when it can mark a repeat.
+
+    ``CaptureSource.read_latest`` returns the newest frame from its background
+    reader; if that reader has not produced another frame yet, the *same* one
+    comes back. Its ``status.last_seq`` advances once per delivered frame, so an
+    unchanged counter means "no new frame". Only a source with a running
+    background reader has that marker: a video file is read synchronously and
+    every call decodes the next frame, so its counter says nothing about repeats
+    and is ignored (``None``).
+    """
+    source = getattr(sensor, "capture_source", None)
+    if source is None:
+        source = getattr(sensor, "_capture_source", None)
+    if source is None or not bool(getattr(source, "running", False)):
+        return None
+    status = getattr(source, "status", None)
+    try:
+        return int(getattr(status, "last_seq"))
+    except (TypeError, ValueError):
+        return None
+
+
 def process_chain(sensor) -> tuple[Optional[np.ndarray], str]:
    """Run the live processing chain for one already-read frame.

@@ -595,6 +623,9 @@ class SdkAcquisitionWorker(RecordSinkMixin):
        saw_frame = False
        last_frame_mono = 0.0
        none_streak_start: Optional[float] = None
+        #: The capture sequence of the last frame that was actually *processed*:
+        #: an unchanged counter means the same frame came back again.
+        last_delivered: Optional[int] = None
        # Per-session monotonic frame index: the join key shared by the video,
        # the CSV and the calibration records.
        frame_id = 0
@@ -615,6 +646,18 @@ class SdkAcquisitionWorker(RecordSinkMixin):
                            break

                t0 = time.perf_counter()
+                # ``CaptureSource.read_latest`` hands back the newest frame and,
+                # until the pipeline produces another one, that **same** frame
+                # again. Treating the repeat as a new sample doubled every
+                # measurement (a 30 fps device recorded 63 rows per second) and
+                # made the reported rate this loop's own polling rate. A source
+                # without a background reader (a video file) decodes a new frame
+                # per call, so it has no repeat marker and is left alone.
+                delivered = _capture_sequence(sensor)
+                if delivered is not None and delivered == last_delivered:
+                    time.sleep(CAPTURE_REPEAT_POLL_S)
+                    continue
+                last_delivered = delivered
                processed = sensor.get_img()
                if processed is None:
                    if saw_frame and is_file and none_streak_start is None:
@@ -658,9 +701,14 @@ class SdkAcquisitionWorker(RecordSinkMixin):
                compute_error = compute_error or chain_error
                t1 = time.perf_counter()
                frame_times.append(t1)
-                cutoff = t1 - 1.0
+                cutoff = t1 - FPS_WINDOW_S
                frame_times = [t for t in frame_times if t >= cutoff]
-                fps = float(len(frame_times)) if frame_times else 0.0
+                # A *rate*, not a count: counting arrivals in a one-second window
+                # reads one frame high whenever the window holds a full second of
+                # a jittery stream. The span between the first and last arrival is
+                # exactly what "frames per second" means.
+                span = frame_times[-1] - frame_times[0] if len(frame_times) > 1 else 0.0
+                fps = (len(frame_times) - 1) / span if span > 0.0 else 0.0
                forces = read_forces_from_sensor(sensor, config_ref=mode.config_ref)
                invalid_reason = forces.get("invalid_reason") or compute_error
                raw_frame = _ensure_bgr(raw)