Skip to content
Back to Home
EN한글
In progress

Clear Box / DashPi

Offline-first Raspberry Pi accident-analysis system

In-progress capstone/exploration: Raspberry Pi 5 + Camera Module captures incident windows, runs YOLO + ByteTrack, and builds an evidence timeline/report. Legal fault judgment is out of scope.

View Code

Overview

My role
Team contributor / product & systems exploration

Operational context

Conventional dashcams store video but leave distressed drivers to interpret raw footage alone.

Users & stakeholders

  • Drivers needing structured post-incident information (target users)

Problem

Post-accident response needs structured evidence timelines—not only raw video—and must avoid pretending to decide legal fault.

Constraints

  • Offline-first / edge hardware limits
  • Explicitly out of scope: legal fault determination
  • Still in progress — not a production deployment claim

My ownership

  • Exploring on-device capture, detection/tracking, and report structuring

Solution

Prototype pipeline: ring-buffer capture → event trigger → YOLO detection → ByteTrack → timeline/report generation on Raspberry Pi.

Key decisions

Assistance report, not verdict

Overclaiming legal judgment would be harmful and inaccurate.

Position as evidence structuring and response assistance only.

Trade-off: Narrower marketing claim; clearer ethical boundary.

Measured impact

WIP
In progress — not production impact
Individual ownership

Technologies

Raspberry Pi 5Camera Module 3YOLOByteTrackPython

Evidence links