Live Pipeline Telemetry

Real-time object detection and tracking analytics

Active Detections 0
Violations Logged 0
Processing Speed 25.0 FPS
Real-time Surveillance Stream
ACTIVE CV DETECTOR

Upload Traffic Video

Drag & drop or click to upload MP4/AVI files

Choose File
Pipeline Parameters
REAL-TIME DETECTOR TELEMETRY LOG FLOW ACTIVE
[SYSTEM] SpeedGuard Vision Engine initialized and standing by.
[SYSTEM] Upload a traffic video or trigger the synthetic simulator to begin tracking.
Oblique Road Plane Calibration Drag anchors to map road perspective grid
Drag anchors to match the lane boundary corners.
Perspective Homography Equations

1. Oblique-to-BEV Homography Transformation

To estimate linear speeds correctly, we construct a 3x3 projective transformation matrix $H$ mapping oblique camera pixels $(x,y)$ to flat bird's-eye coordinates $(X_w, Y_w)$:

Xw
Yw
1
  =  H  × 
x
y
1
  = 
h00h01h02
h10h11h12
h20h21h22
  × 
x
y
1

2. Real-time Homography Matrix $H$

Recomputed dynamically based on your drag anchor points:

[ 1.423, -0.841, 120.4 ]
[ 0.124, 1.962, -80.6 ]
[ 0.000, -0.002, 1.0 ]

3. Average Speed Calculation

Once bounded boxes cross Virtual Line A (entry) and Virtual Line B (exit), the distance traveled is projected using $H$, and speed is estimated as:

Speed (km/h) = 3.6 × Distance (meters) Travel Time Δt (seconds)
Δt = Δframes / FPS
Image Anchors: [(160,480), (480,480), (380,220), (260,220)]
Computed Pixel Scale: 0.0357 meters/pixel
Vehicle Speed Distribution
Classification Breakdown
Violations Database Log
Timestamp Vehicle ID Class Speed Limit Excess License OCR Status
No speed violations logged yet. Trigger the demo or upload a video!