XCORE-VISION
Real-time embedded vision.
No OS, No cloud. No FPGA complexity.
A compact vision platform powered by the XCORE.AI multicore architecture. Run AI models like MobileNet V2 directly on the device with deterministic, low-power performance.
Integrated camera
Ready to go out the box
MobileNet V2
Run AI models on-device
Low power
Designed for edge devices
Deterministic
Real-time, predictable performance
Parallel processing
Multiple tasks at once
No cloud
Secure, private and offline
XMOS Labs project overview
See what you can do with XCORE-VISION
A tour of the platform – real-time at the edge, running MobileNet V2 locally on the XCORE.AI device. No cloud. No OS. Just deterministic, low-power vision.
What can you build?
Face recognition
Secure access and identification.
Object detection
Recognise, classify and count objects.
Industrial inspection
Detect defects and ensure quality
Robotics
Enable vision for autonomous objects
Presence detection
Monitor occupancy and movement
See more ideas and community projects ->
Everything you need
Integrated camera module
XCORE-VISION board
lib_camera library
MobileNet V2 examples
Sample Applications
Documentation and Guides
Powered by XCORE.AI
Deterministic multicore architecture
Up to 100 GOPS (INT8) AI performance
Hardware-level parallelism for real-time systems
Low-power with advanced power states
