Architecture
From capture to intelligence to application
A layered architecture: each layer has a clear responsibility, is independently replaceable, and exposes standardized interfaces to the layer above.
Processing Pipeline
End-to-end pipeline
The platform does not bind to any specific algorithm — algorithms exist as plugins, and applications consume only standardized vision features.
Input
- Vision Input
- RTSP · ONVIF · GB28181
- USB · HDMI · File
Processing
- Detection
- Tracking
- Recognition
Output
- REST API
- Webhook
- Events
Closed-loop Vision
From passively receiving images to actively acquiring better ones
In traditional systems AI can only passively accept images. Bang Vision forms a loop through Control + Optics + Sensor: understand → control → acquire better data → understand again.
Understand
- AI Analysis
- Vision Features
Control
- Lens · Sensor · ISP
- Exposure · Focus · Zoom
Acquire
- Better Image
- Richer Information
Deployment
Deployment Targets
We are upfront about what is not supported — focusing on real-time edge vision processing is what makes the platform strong.
Current Support
Android Edge Devices
RTSP input and on-device AI inference.
Linux Edge Devices
ARM64 with GPU/NPU acceleration (hardware-dependent).
RK3588 Platforms
Evaluation and development support.
In Progress
Industrial Cameras
Custom Sensor Integration
Hardware Control Layer
Get Started
See the platform in action
Explore the technical demo, or tell us about your use case to request a live walkthrough.