Perception Kits · Video Analytics Kit
HyperEdge M300F – Edge AI Video Analytics Kit (Standard)
The HyperEdge M300F Video Analytics Kit (Standard) is a pre-matched edge AI hardware kit for dual-zone, on-premise video analytics. It pairs CORE BOLT's HyperEdge M300F Edge AI Computer (NVIDIA Jetson Orin Nano or Orin NX, up to 157 TOPS) with two SENSING SGigE-E50-H2MC 5 MP GigE Vision cameras on independent Ethernet ports, for person detection, people counting and occupancy monitoring that keeps video on site.
Specifications
Technical Specifications
HyperEdge M300F — Video Analytics Kit (Standard) — System Configuration
| Parameter | Specification |
|---|---|
| Product Type | Integrated AI Perception Kit |
| Kit Code | CB-CCTV-705 |
| Primary Compute | CoreBolt HyperEdge M300F |
| Compute Configuration | NVIDIA Jetson Orin Nano / Orin NX |
| Maximum AI Performance | Up to 157 TOPS |
| 3D LiDAR | Not required for this application |
| Cameras / Vision | 2× SENSING SGigE-E50-H2MC |
| Sensor Interfaces | GigE Vision (RJ45) |
| Perception Functions | Two independent GigE ports for dual-zone coverage; 5 MP resolution vs. the Essential kit's 1.3 MP; PoE-optional ports for third-party IP cameras |
| Deployment Environment | Indoor retail, offices, transport hubs; small multi-zone sites |
| Typical Applications | Dual-zone on-premise video analytics |
HyperEdge M300F — AI Edge Computer
| Parameter | Orin Nano 4gb | Orin Nano 8gb | Orin Nx 8gb | Orin Nx 16gb |
|---|---|---|---|---|
| AI Performance | 20 TOPS | 40 TOPS | 70 TOPS | 100 TOPS |
| AI Performance — Super Mode | 34 TOPS | 67 TOPS | 117 TOPS | 157 TOPS |
| GPU | 512-core NVIDIA Ampere GPU, 16 Tensor Cores | 1024-core NVIDIA Ampere GPU, 32 Tensor Cores | 1024-core NVIDIA Ampere GPU, 32 Tensor Cores | 1024-core NVIDIA Ampere GPU, 32 Tensor Cores |
| CPU | 6-core Arm Cortex-A78AE | 6-core Arm Cortex-A78AE | 6-core Arm Cortex-A78AE | 8-core Arm Cortex-A78AE |
| Memory | 4GB 64-bit LPDDR5 34 GB/s | 8GB 128-bit LPDDR5 68 GB/s | 8GB 128-bit LPDDR5 102.4 GB/s | 16GB 128-bit LPDDR5 102.4 GB/s |
HyperEdge M300F — Interfaces & Environment
| Parameter | Specification |
|---|---|
| Ethernet | 2 × RJ45 Gigabit Ethernet, independent IP, PoE optional |
| USB | 2 × USB 3.2 Type-A, 1 × Type-C debug, 1 × Type-C OTG |
| Serial / CAN / GPIO | 1 × RS-232, 1 × RS-485, 1 × CAN-FD (DB9), 4 × GPIO |
| Expansion | Mini PCIe (4G LTE + SIM), M.2 2280 Key M (NVMe SSD), M.2 2230 Key E (Wi-Fi / BT) |
| Power Input | 9–24 V DC (DC barrel) |
| Operating Temperature | -20°C to +70°C |
| Cooling | Active fan cooling, rugged aluminium enclosure |
| Dimensions | 161 × 100 × 50 mm |
SENSING SGigE-E50-H2MC — 5 MP GigE Camera
| Parameter | Specification |
|---|---|
| Sensor | HAITU HT50A, 5 MP, 2/3", mono |
| Resolution | 2448 × 2048 |
| Frame Rate | 23.5 fps |
| Shutter | Rolling |
| Lens Mount | C-mount (lens not included) |
| Interface | GigE Vision, 1000 Mbit/s, RJ45, up to 100 m |
| Power | 12 V DC ± 3 V (separate supply, no PoE) |
| Operating Temperature | 0°C to +50°C |
| Weight | 51 g |
Integration & Connectivity
| Component | Interface | Connected To | Function |
|---|---|---|---|
| 2× SENSING SGigE-E50-H2MC | GigE Vision (RJ45) | HyperEdge Gigabit Ethernet | Dual-zone person detection & occupancy tracking |
| HyperEdge M300F | NVIDIA Jetson Orin | System Compute | AI Inference & Sensor Processing |
| NVMe SSD | M.2 Key M | HyperEdge M300F | Models, Applications & Data Storage |
Deployment Notes
| Parameter | Details |
|---|---|
| Customer-Supplied Items | Network switch (beyond 2 cameras), NTP time source, VMS / dashboard endpoint, camera lenses and 12 V supplies |
| Port Budget | Both M300F Gigabit Ethernet ports are used by the two cameras — add a switch for more |
| Camera Power | The SENSING GigE cameras need separate 12 V DC supplies; the M300F's optional PoE only powers third-party PoE cameras |
| Stream Sizing | Video decode capacity differs between Orin Nano and Orin NX — size the module to the number of streams before ordering |
Overview
Dual-Zone, On-Premise Video Analytics at the Edge
Discover the HyperEdge M300F edge AI video analytics kit with Jetson Orin and two 5 MP GigE cameras for on-premise people counting and occupancy monitoring.
Retail stores, offices and transport hubs increasingly want occupancy and footfall data without streaming video to the cloud. The HyperEdge M300F processes video locally on an NVIDIA Jetson Orin module, so footage can stay on the premises and analytics continue without an internet connection. Two independent-IP Gigabit Ethernet ports give each camera its own zone.
Compute is available in four configurations: Orin Nano 4GB (34 TOPS), Orin Nano 8GB (67 TOPS), Orin NX 8GB (117 TOPS) and Orin NX 16GB (157 TOPS), all in Super Mode. The SENSING cameras use a 5 MP 2/3" monochrome sensor (2448 × 2048, 23.5 fps) with C-mount optics. This is the mid-tier Standard build in CORE BOLT's video analytics kit family, a step up from single-feed retrofit hardware.
The kit is supplied as hardware. Video analytics applications are developed or provided by the customer.
What's in the Kit
- 1 × CORE BOLT HyperEdge M300F Edge AI Computer (NVIDIA Jetson Orin Nano or Orin NX)
- 2 × SENSING SGigE-E50-H2MC 5 MP GigE Vision cameras (lenses not included)
- Kit code: CB-CCTV-705
Key features
- Up to 157 TOPS (Orin NX 16GB, Super Mode), with Orin Nano and Orin NX options
- 2 × RJ45 Gigabit Ethernet with independent IP addresses, PoE optional for third-party IP cameras
- 5 MP GigE Vision cameras, 2448 × 2048, C-mount, cable runs up to 100 m
- RS-232, RS-485, CAN-FD and 4 × GPIO for building and access-control integration
- Compact 161 × 100 × 50 mm rugged aluminium enclosure, -20 °C to +70 °C, 9–24 V DC
- 4G LTE (Mini PCIe), Wi-Fi/Bluetooth (M.2 Key E) and NVMe SSD expansion
Applications
Video analytics, smart buildings, smart retail and digital signage, smart transportation hubs, smart infrastructure, and occupancy and footfall monitoring in offices and small multi-zone sites.
Frequently Asked Questions
What is an edge AI video analytics kit?
It is a pre-matched set of cameras and an edge AI computer that runs video analytics models on site, rather than sending video to the cloud for processing.
Can AI video analytics work without an internet connection?
Yes. On an edge AI computer such as the HyperEdge M300F, video is processed locally, so analytics can run without a cloud connection. Footage can also remain on the premises.
How many cameras does the HyperEdge M300F kit support?
The kit includes two GigE cameras on two independent Ethernet ports. More cameras need a network switch, and the number of streams depends on the Orin Nano or Orin NX module selected. For eight cameras, see the HyperEdge 700R Video Analytics Kit (Advanced).
Should I choose Jetson Orin Nano or Orin NX for video analytics?
Orin Nano (34–67 TOPS) suits fewer streams and lighter models. Orin NX (117–157 TOPS) offers more AI performance and video decode capacity for more streams or heavier models.
Are monochrome cameras suitable for people counting?
Yes. Person detection and counting rely on shape and motion, so monochrome cameras work well and provide good sensitivity. Colour cameras are only needed if the application uses colour information.
What extra items are needed to install the kit?
Camera lenses, 12 V DC power supplies for the cameras, an NTP time source, a VMS or dashboard endpoint, and a network switch if you add more than two cameras. Request a perception kit to discuss your site.