Small Retailer
Centroid-64
Centroid-64 is engineered with clearly defined processing limits, input standards, and output formats. Each specification reflects real deployment constraints, ensuring predictable performance in production environments.
Category |
Specification |
|
Dimensions |
419 (W) x 320 (D) x 68 (H) |
|
Weight |
Device: 5.4 kg |
|
Casing |
Anodised Aluminium Casing |
|
CPU |
Quad Cortex®‑A76 @ 2.2~2.4GHz and Quad Cortex®‑A55 @ 1.8GHz |
|
GPU |
Arm Mali™ G610MP4 |
|
NPU |
48 TOPS |
|
No. of AI Channels |
64 |
|
Input Power |
AC 110~230V, 50~60Hz |
|
Network Interface |
Ethernet |
|
Processor Class |
Embedded multi-core edge processor |
Video Input
Supports RTSP and ONVIF-compatible IP cameras. Compatible with standard network video streams from existing CCTV infrastructure. No proprietary encoding or camera dependency required.
Attribute |
Summary |
|
Supported Protocols |
RTSP / ONVIF |
Processing Capacity
Supports up to 64 concurrent video streams per unit. Streams are processed continuously in real time with parallel execution across all active inputs. No batch processing or deferred workloads.
Attribute |
Summary |
|
Maximum Streams |
64 |
|
Processing Mode |
Real-time |
|
Execution |
Parallel |
Analytics Output
Outputs structured data via API and data streaming interfaces. Data is formatted for integration with dashboards, BI platforms, or external systems.
Output Data
In / Out Counts
Zone Occupancy
Movement Flow
Queue Indicators
System Architecture
Operates as a self-contained edge appliance. All video processing is performed locally. No dependency on external GPU servers or cloud-based processing.
Edge Appliance
Local Processing
No Cloud Dependency
Performance Characteristics
Designed for continuous operation under full load. Maintains consistent processing behaviour across all active streams within defined capacity. No dynamic scaling or workload redistribution.
Continuous Operation
Stable Throughput
Fixed Capacity Behaviour
Power and Physical
Designed for deployment in standard IT environments. Compact form factor suitable for comm rooms or back-office installation. Lower power consumption compared to GPU-based compute systems.
Attribute |
Summary |
|
Form Factor |
TBD |
|
Power Consumption |
TBD |
|
Mounting |
Rack / Shelf |
Privacy and Compliance
Processes video data locally. No facial recognition or identity tracking. Exports anonymised analytics data only. Supports compliance with GDPR and similar data protection frameworks.
No Facial Recognition
No Identity Tracking
GDPR-Aligned