Dmood176Final

Project Overview (Source: Dualmood)

Dualmood builds a security camera using a Raspberry Pi, primarily to save money and minimize power consumption and wiring.

Cost & Hardware Advantages

• Spent < $50 on materials.
• Compared to commercial security cameras costing ~$200–$400 per camera.
• Minimal wiring:
  • Powers the camera via USB-C (cheap and convenient).
• Low power usage:
  • About ~3.5 W average, compared to ~20 W+ for many professional cameras.

Networking Setup & Remote Access (Major Technical Effort)

• Uses IPv4 forwarding and also experimented with some IPv6.
  • The IPv4/IPv6 approach became problematic.
• Steps taken:
  • Created a new subnet.
  • Set up a VPN to tunnel traffic between a personal computer and the Raspberry Pi for remote access.
• Issues encountered:
  • DNS issues during VPN setup.
• VPN/tooling choices:
  • Tried Twingate first (previously reliable), but encountered a critical problem:
    • When active on both devices, there was no internet access.
  • Switched to Tailscale, which worked smoothly:
    • Resolved earlier VPN/DNS problems and improved connectivity.
• Additional friction:
  • Connecting from the Pi to the file server required re-authentication each time (the Pi has no screen).
  • Adjusted workflow to resolve this, aided by the shift to Tailscale.

Lesson learned: networking and remote access issues can be time-consuming and are often unavoidable.

Camera Software, Installation & Configuration

• Followed GitHub instructions and installed MotionEye.
• Configured:
  • Camera definition
  • Resolution and FPS
  • Video exposure/brighting behavior
    • Auto brightness did not work well, so exposure/brightness were adjusted manually
  • Local file save location
  • File server target for uploads (described as relatively straightforward)

File Storage & Backups (Data Management)

• After recording, videos must be stored on a drive:
  • Determined where recordings are stored
  • Moved copies to a cloud server
  • Configured a RAID partition on the server for large and growing storage needs (video grows quickly)

SFTP Plugin Issue & Custom Backup Script

• The built-in SFTP plugin in the camera system did not work.
• Instead, a custom backup script was written to copy locally stored video files to the cloud server.
• The script was scheduled using Linux cron to run nightly.

Scripting & Automation

• Created a custom script to perform backups because the SFTP plugin failed.
• Added it to crontab for regular (nightly) execution.

I/O and Motion Detection Behavior

• Motion detection requires processing the input stream to detect movement.
• Used motion detection masking:
  • Example: masked out a sidewalk area to prevent unwanted triggers.
• Video playback/UI supports:
  • Live streaming
  • Viewing past pictures/videos
  • Sorting by date

Admin vs. User Configuration (UI Observations)

• Primary user:
  • Limited settings (e.g., number of streams visible)
• Admin user:
  • Full settings access, including:
    • Brightness adjustment
    • Resolution
    • Rotate
    • FPS
    • File storage details
    • Disk usage

Troubleshooting Lessons Learned

• VPN/network connectivity can take much longer than expected; networking issues are inevitable.
• Changing VPN/device configurations can lead to getting locked out of the Pi.
• Vendor integrations can be too rigid:
  • Twingate wasn’t designed for non-user-facing devices, contributing to the switch.
  • MotionEye/provided plugins were tightly integrated and lacked flexibility.
• Recommendation for future work:
  • Consider writing more components from scratch if it better matches requirements.
• Always plan for storage growth, since video accumulates rapidly.

Main Speaker / Source

Dualmood (presenter/author)