Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/Xander44-4/traffic_reducer/llms.txt

Use this file to discover all available pages before exploring further.

Live mode connects the YOLOv8 detection pipeline directly to the signal decision logic, updating the green-light assignment every 2 seconds based on real vehicle counts read from the active video source. No manual slider input is needed — the camera does the counting.

​
Activating live mode

1

Select a video source

Ensure a video source is active. If the dashboard shows ● INACTIVO, click EN VIVO or VIDEO LOCAL in the camera feed card header, or make a selection in the source modal. The stats polling interval will start automatically once a source is chosen.
2

Enable the toggle or click the navbar button

Toggle the Simulación en vivo switch in the hero section to the ON position, or click the Activar Control button in the navigation bar. Both controls are synchronized — either one flips the same internal cvModeActive flag.
3

Confirm the mode change

The four manual range sliders are disabled immediately. The navbar button label changes to Desactivar Control. The intersection graphic begins updating from camera-reported counts instead of slider values.

​
What happens during live mode

Once live mode is active, the frontend runs two independent polling loops:

​
Prediction loop — every 2 000 ms

The frontend calls POST /predict with { "live_mode": true } every 2 seconds: 
// script.js — startLiveLoop()
liveInterval = setInterval(updatePrediction, 2000);
The server reads the latest per-zone vehicle counts from the YOLO pipeline: 
# app.py — /predict route
counts = camera.get_counts()
norte = float(counts['norte'])
sur   = float(counts['sur'])
este  = float(counts['este'])
oeste = float(counts['oeste'])

result = int(np.argmax([norte, sur, este, oeste]))
camera.set_phase(result)   # updates internal phase state
The winning direction index (0 = NORTE, 1 = SUR, 2 = ESTE, 3 = OESTE) is returned in the prediction field. The dashboard updates the intersection graphic: the winning direction’s green bulb activates, all others revert to red, and the Prioridad activa card shows the direction name. A typical live-mode prediction response: 
{
  "prediction": 2,
  "traffic_data": {
    "norte": 3.0,
    "sur": 7.0,
    "este": 14.0,
    "oeste": 5.0
  }
}

​
Stats loop — every 1 500 ms

An independent interval polls GET /stats every 1 500 ms throughout the session (not just in live mode). It updates the sidebar counts (N, S, E, O), pedestrian count, emergency status, and the AI Decision phase label: 
// script.js — fetchStats()
statsInterval = setInterval(fetchStats, 1500);
A sample /stats response: 
{
  "norte": 3,
  "sur": 7,
  "este": 14,
  "oeste": 5,
  "pedestrians": 2,
  "emergency": false,
  "phase": "ESTE",
  "priority": "ESTE",
  "priority_idx": 2,
  "mode": "youtube"
}

​
Emergency override

If the camera detects an emergency vehicle, camera.get_counts() sets emergency: true. When this reaches the dashboard through either the /predict or /stats response:
  • The Prioridad activa card displays EMERGENCIA in #b64400 (amber-red).
  • The sidebar Emergencia field changes from NO to SÍ in the same colour.
  • The manual sliders remain locked (they were already disabled in live mode).
The emergency display is cleared automatically on the next poll once the emergency vehicle is no longer detected.

​
Deactivating live mode

Toggle Simulación en vivo back to OFF, or click Desactivar Control in the navbar. The frontend:
  1. Clears the 2 000 ms prediction interval (clearInterval(liveInterval)).
  2. Re-enables all four range sliders.
  3. Fires one immediate POST /predict call with the current slider values so the intersection graphic reflects manual state instantly.
The stats polling loop at /stats continues running independently — it is not tied to the live-mode toggle.
Live mode works fully with the local video source — no internet connection is required. This makes it useful for offline development and integration testing. Set LOCAL_VIDEO_SPEED to 1.0 in app.py for real-time simulation, or a higher value such as 2.0 for accelerated scenario testing.

Build docs developers (and LLMs) love

Get started for freeTalk to us