Anyone done live RTSP streaming + AI detection in Bubble?
Hey all,
I’m working on a personal project that has live camera and AI detection but the two features have me looking for war stories before I commit to an architecture:
1. Live RTSP camera streaming
Need to display live IP camera feeds in a dashboard. Since browsers can’t play RTSP natively, I know I need a streaming service in front of it. Currently weighing:
• Mux (RTSP ingest, mature web player)
• Ant Media Server (self-hosted or cloud, RTSP native)
• AWS Kinesis Video Streams (RTSP-friendly but enterprise SDK)
• AWS IVS (would need an RTSP → RTMPS relay)
If you’ve built this in Bubble, I’d love to know:
• Which service you went with and why
• How you embedded the player (HTML element? Plugin? Custom JS?)
• How you handled multi-tenant playback auth (so Company A can’t watch Company B’s streams)
• Latency you achieved
• Monthly infra cost per camera/stream
2. AI detection on those feeds
Want motion / person / after-hours detection running on the same streams, auto-creating alerts in the app. Considering:
• AWS Rekognition Video Streams
• OpenAI Vision (frame-based)
• Roboflow
• Self-hosted model via API
Curious about:
• Cost vs. accuracy trade-offs in your experience
• Frame-extraction pipeline (continuous? scheduled? event-triggered?)
• How bad false positives got and what you did about them
If you’ve shipped either of these in production on Bubble — even partially — I’d really value hearing what worked and what didn’t. Happy to share back what I learn from the build.
Thanks!