Mobile Use and Seat Belt Compliance

Smart Solutions for Most Pressing Safety Issues

Don't Text and Drive

Use of mobile phone while driving not only puts the driver at risk, but is a threat to all other road users too, including pedestrians, cyclists, and other vehicles. Reaction times reduce by half, and risk of crash can increase up to 23 times, making distracted driving one of the most risky actions a driver can take.

Buckle Up

Wearing a seat-belt reduces the risk of death as well as serious injury in a crash by about half. Three-point seat-belt is the most important passive safety device in a vehicle. Yet, seat-belt compliance is low in many countries.

Smart Enforcement by Image Analytics

Hazen.ai’s seat-belt and mobile phone sensing software is built using the latest Deep Learning algorithms, analyzing images in real-time and generating active triggers on violations. High accuracy of our software minimizes false alarms and provides actionable intelligence in real-time.

Low-cost and high-accuracy enforcement saves lives. Every vehicle passing by the camera is scanned by our efficient software. The high likelihood of being issued a citation changes driver’s behavior towards safer practices, making our roads safer.

Outstanding Performance

>90% accuracy

Field trials and case studies have shown accuracy exceeding 90% after model adaption and training. Accuracy measurement is done against human ground-trhuth performance on same images. Accuracy may vary by camera angle and image quality.

Simple and Hassle Free Installation

Works with standard CCTV cameras. Up-to 3 lane coverage with 1920 x 1080 resolution sensor.

Options for both Edge-based and On-Prem deployments.

Road-side installation allowed (rather than gantry requirement), making it easy to deploy.

Compatible with both Nvidia and Intel compute hardware.

Real-time performance

Hardware Requirements

Camera Specs

  • Image Input: minimum 2K resolution (1920 x 1080) for 3 lane coverage
  • 8, 10, 0r 12 bit color mode
  • Still image capture mode
  • External IR illumination for night-time performance

Processor Options

  • Compatible with both Intel and Nvidia compute hardware
  • 15fps with Nvidia Xavier NX GPU at edge.
  • 5fps on Intel i7-9900K / 8GB
  • 0.5fps on Intel Atom 3845 + NCS2 VPU

Communication

LTE module required for violation package communication, system health monitoring for edge deployments

Software Specifications

Accuracy

  • Human-grade classification results
  • Learning based system adapts to new environments
  • Accuracy >90% after model adaption

Coverage

  • Up to 3 lanes with single camera on road-side installation at 4 – 6m height.
  • Up to 4 lanes with gantry mounting.
  • Distance-to-vehicle < 20m

Deployment Options

  • On-Edge or On-Premises Options
  • MQTT broker based integration with client’s violation package system (default)
  • More integration options available on request

Sample Results