214
peak INT8 TOPS
on the Metis AIPU
16
camera streams
at 1080p, 20 FPS
8
hands-on projects
to build and run
The Starter Kit
A complete
AI computer for your lab
Built on an Axelera single-board computer with a quad-core Metis AI processing unit. No separate host PC or GPU is needed: students plug in cameras and start building. It ships with the WG DeepInsight software, example models and a starter manual.
End-to-end Edge AI workflow
From setup to real-time inference in 7 steps
1. Set up the device
Power up the Starter Kit and deploy the pre-built WGTech Docker container.
2. Connect cameras
USB cameras, IP/RTSP cameras or recorded video, with live ingestion.
3. Annotate data
Complimentary annotation tool to label frames and export datasets.
4. Train the model
On AWS or student-accessible NVIDIA GPU farms.
5. Deploy to the edge
Your own model or the default one, optimised for the Metis AIPU.
6. See it run live
Real-time bounding boxes and labels, accelerated in hardware.
7. Record and replay
Save streams with inference metadata for validation and analysis.
Learn by building
8 hands-on projects students can try
Real-time object detection
People, chairs, bottles, bags and laptops in a live feed.
People detection & countin
Count people in a classroom, lab or corridor.
Safety helmet detection
Spot whether a safety helmet is worn, using a trained model.
Vehicle detection & parking
Count cars and motorcycles and estimate occupancy.
Image classification
Sort bottles, paper, metal parts or types of leaves.
Visual quality inspection
Tell acceptable objects from defective ones.
Line-crossing counter
Count people or objects crossing a virtual line.
Mini AI surveillance dashboard
Live video with detections, counts, timestamps and FPS.

WG DeepInsight dashboard
Cameras, live view, playback, annotations and models in one interface, so students focus on the AI, not the plumbing.
Why colleges choose it
Real-world skills
Practical experience with edge AI deployment and model optimisation.
Enterprise technology
Professional-grade hardware usually reserved for commercial deployments.
Project-based learning
Ideal for senior design projects and undergraduate research.
What's in the box
Complete hardware
The Axelera SBC Starter Kit, pre-assembled and ready to use.
Starter manual
Step-by-step installation and deployment documentation.
Example models
Pre-configured vision models for immediate testing and learning.
Direct support
Developer documentation and technical support resources.
Ready to Build Your Next Vision AI System?
To request a Technical Demo, please fill the form.
Let's Work
Together
+91 63620 37962
support@wgtech.ai
WG Tech, Bengaluru
WG Tech Solutions Pvt Ltd
Rainmakers Workspaces, No. 759, 3rd Floor, 8th Main Rd, KSRTC Layout, 3rd Phase, J. P. Nagar, Bengaluru,
Karnataka 560078
+1 (408) 391-2142
support@wgtech.ai
WG Tech, Delaware
WG Tech Solutions Inc,
8, The Green, #11166, Dover,
DE 19901, USA
Disclaimer: All images and video content displayed on this website are for representational and illustrative purposes only. Actual software interfaces and hardware performance may vary. For a live demonstration or to view production-grade output, please write to us at support@wgtech.ai to schedule a demo.
Quad-core Metis AIPU
214 peak INT8 TOPS for dedicated AI inference.
8-core Arm CPU
Cortex-A76/A55 host processor on the same board.
Generous memory
16 GB CPU memory, plus dedicated AIPU memory.
64 GB eMMC storage
Room for models, datasets and recordings on board.
Up to 16 camera streams
1080p at 20 FPS, from USB and network/RTSP cameras.
Axelera Voyager SDK
Industry-grade toolchain to compile and deploy vision models.
WG DeepInsight software
Dashboard, annotation tool and pre-built Docker runtime.
Low power
Typical board power of just
9–20 W.


