Hey everyone,
I wanted to share an on-premises software project I’ve been developing called Lot Vulture and get some feedback from people running real-world outdoor cameras.
The software detects parking spot occupancy using AI models. It runs 100% locally on-prem, connects to existing RTSP camera streams, and provides real-time data that can be integrated into other systems to tell you if parking spots are occupied or not occupied. It could be used for a home driveway with 1 or 2 spots or a commercial parking lot with thousands of spaces.
What it does:
• Interactive Visual Editor: You pull up the web UI, view your camera's live snapshot/stream, and draw 4-point polygon zones over parking spots/driveway spaces.
• Real-Time Status: Evaluates spaces in real-time and streams occupancy status (Vacant / Occupied) over WebSockets to a live dashboard.
• Dual-Threshold Hysteresis: Built-in dampening logic so transient shadows, headlights, rain, or a person walking across a spot don't cause notification flicker.
• Visual Timeline Scrubbing: Rewind and scrub through the day minute-by-minute with snapshots for every state transition.
• Alerting / Integrations: Supports MQTT and Webhooks (ideal if you want to push occupancy status to Home Assistant, trigger relays/signage, or send alerts when the driveway/lot is full).
Hardware & Performance:
• Native Windows Installer (.exe): Packages the dashboard and C++ inference engine. No Docker, WSL, or external Python environments required on
Windows.
• Linux / Docker: Also available as a container .
• Can run against camera substreams to keep resource consumption low.
It's source-available, and the Community Edition is completely free to deploy and use. I also am releasing a commercial version that has improved accuracy AI models and GPU support.
Screenshots & Demo:
GitHub / Releases:
GitHub - lotvulture/lotvulture: Real-time parking lot occupancy detection using existing security cameras and on-device computer vision.
(Windows installer LotVulture_Setup_*.exe is under the Releases tab)
I would love to hear your thoughts, test results with different camera angles/focal lengths, or feature requests. I imagine there are many use-cases that I can't think of, but would be great to build into the software to make it easier beyond simple MQTT/WebHook style integrations.
I wanted to share an on-premises software project I’ve been developing called Lot Vulture and get some feedback from people running real-world outdoor cameras.
The software detects parking spot occupancy using AI models. It runs 100% locally on-prem, connects to existing RTSP camera streams, and provides real-time data that can be integrated into other systems to tell you if parking spots are occupied or not occupied. It could be used for a home driveway with 1 or 2 spots or a commercial parking lot with thousands of spaces.
What it does:
• Interactive Visual Editor: You pull up the web UI, view your camera's live snapshot/stream, and draw 4-point polygon zones over parking spots/driveway spaces.
• Real-Time Status: Evaluates spaces in real-time and streams occupancy status (Vacant / Occupied) over WebSockets to a live dashboard.
• Dual-Threshold Hysteresis: Built-in dampening logic so transient shadows, headlights, rain, or a person walking across a spot don't cause notification flicker.
• Visual Timeline Scrubbing: Rewind and scrub through the day minute-by-minute with snapshots for every state transition.
• Alerting / Integrations: Supports MQTT and Webhooks (ideal if you want to push occupancy status to Home Assistant, trigger relays/signage, or send alerts when the driveway/lot is full).
Hardware & Performance:
• Native Windows Installer (.exe): Packages the dashboard and C++ inference engine. No Docker, WSL, or external Python environments required on
Windows.
• Linux / Docker: Also available as a container .
• Can run against camera substreams to keep resource consumption low.
It's source-available, and the Community Edition is completely free to deploy and use. I also am releasing a commercial version that has improved accuracy AI models and GPU support.
Screenshots & Demo:
GitHub / Releases:
GitHub - lotvulture/lotvulture: Real-time parking lot occupancy detection using existing security cameras and on-device computer vision.
(Windows installer LotVulture_Setup_*.exe is under the Releases tab)
I would love to hear your thoughts, test results with different camera angles/focal lengths, or feature requests. I imagine there are many use-cases that I can't think of, but would be great to build into the software to make it easier beyond simple MQTT/WebHook style integrations.