VR-Based Eye Tracking Sobriety Test
The system captures real-time eye movement via VR headsets and supports AI-based evaluation of THC-related visual markers during roadside testing.
AI Video Platform for Automated Sobriety Testing
AI-powered marijuana detection device with VR-based eye tracking, automated sobriety test workflows, and secure video processing for compliance.
Project Overview
We partnered with a US-based startup to engineer the backend video infrastructure for an innovative marijuana detection device. The goal was to automate roadside sobriety test workflows by replicating and enhancing traditional Drug Recognition Expert (DRE) methods using VR technology and real-time AI drug testing. Our solution supported secure video processing, audio/subtitle overlays, and seamless cloud integration to improve accuracy, traceability, and scalability. This advanced eye tracking sobriety test system is now actively used in law enforcement and workplace safety protocols to assess cannabis-induced impairment.
Overcoming Roadblocks in Impairment Detection
Subjective and inconsistent manual tests created a pressing need for objective, reliable, and scalable detection systems to ensure safety across law enforcement and workplaces.
Inconsistent and Non-Digital DRE Testing
Manual DRE eye tests lacked objectivity, standardization, and reliable video documentation, creating inconsistencies in impairment assessments during roadside sobriety test procedures.
Lack of Scalable Video Infrastructure
No solution existed to reliably capture, process, and store high-volume eye movement recordings, a requirement for scalable automated sobriety test deployment.
Missing Audio/Subtitles in Recordings
Raw recordings lacked guided voice prompts and synchronized subtitle overlays, critical for regulatory reviews, training consistency, and future rehabilitation app development.
Security and Privacy Gaps
High-sensitivity video data needed encryption, structured metadata tagging, and privacy controls, which were absent in existing video workflows.
Fragmented Metadata and Traceability
No reliable way existed to connect recordings to session metadata like user ID, test type, and timestamps, hindering compliance and usability.
Our Tailored Backend Video Processing Solution
We developed a complete backend workflow to automate video ingestion, processing, audio/subtitle integration, and re-uploading with end-to-end session traceability.
Firebase-Powered Video Automation
Integrated Firebase SDKs to automate video retrieval and uploads based on structured session metadata while ensuring encrypted access and bucket-level control.
Smart Audio & Subtitle Sync
Locally processed videos were enhanced with audio prompts and subtitle overlays by aligning SRT files to specific eye movement cues used in impairment detection.
Structured Upload & Routing Logic
Videos were renamed and routed into pre-defined Firebase folders, ensuring logical storage paths based on test type, user, and timestamp.
Volume Handling & Retry Logic
Processing pipelines supported bulk video handling, download retries, file integrity validation, and optional compression or watermarking for public-facing use.
Session-Centric Metadata Tagging
Each recording was linked to key metadata like officer ID, result, and test type, supporting downstream analytics and regulatory documentation.
Core Features We Delivered
Automated Video Retrieval from Firebase
Session-specific videos are pulled automatically, streamlining the processing workflow and reducing the need for manual intervention.
Audio and Subtitle Overlay Engine
FFMPEG-powered logic injects synchronized audio and subtitle cues, offering a consistent test structure and improved replayability for legal or training use.
Dynamic Re-Upload Workflows
Processed videos are automatically re-uploaded with traceable paths, flag-based visibility (private/public), and expiration logic.
Compliance and Traceability Support
Every session is audit-ready, with encrypted processing, structured logs, and metadata for full visibility into who was tested, when, and how.
Tech Implementation
- FFMPEG
- Python
- 3rd-Party SDK
- encrypted data storage
- SRT
Scaling Cannabis Impairment Detection from Pilot to Production
Our automated pipeline accelerated video readiness, improved compliance, and enabled data-driven expansion into enterprise and law enforcement applications.
Digitized Traditional Drug Recognition Processes
We modernized subjective roadside testing with objective, VR-based eye tracking sobriety test workflows supported by structured video evidence.
Accelerated Processing and Turnaround Time
Audio and subtitle overlay automation reduced session processing time by 70%, enabling quicker decision-making at the point of contact.
AI Drug Testing at Scale
With a scalable video infrastructure, our client deployed an AI drug testing solution capable of handling large volumes of sessions across jurisdictions.
Foundation for Advanced Use Cases
The system now supports additional developments such as rehabilitation app development and pupil dilation analytics using previously recorded sessions.
Improved Compliance and Safety Standards
With secure handling, metadata tagging, and visibility flags, the platform meets the privacy and audit needs of both public and private sector stakeholders.