CM-EMB Embedded Module | Artificial Vision





	CM-1K Neural Network Chip

	CM-IR2K Image Recognition Board

	CM-SK1K Starter Kit

	CM-PM1K Prototyping Module

	CM-EB2K Evaluation Base Board

	CM-EMB Embedded Module

	CM-V1KU Vision Sensor Board

	Contact C4AV

	Forums

	Home

Online Payment Service

CM-EMB Embedded Module

The CogniMem Embedded Module (CM-EMB) is a miniature OEM board populated with one or two CogniMem chips. Its network of 1024 or 2048 neurons implements two renown non-linear classifiers and can recognize patterns at high speed while coping with ill-defined data, unknown events and changes of contexts and working conditions. Communication is made via a parallel bus or serial bus accessible on the 80-pin hirose connector. Up to four modules can be stacked on top of each other thus expanding the size of the network by increments of 1024 or 2048 neurons, up to 8192.

Purchase The CM-EMB1K - CM-EMB2K - CM-EMB2K-I »

Features

	1 or 2 CogniMem chips (i.e. 1024 or 2048 neurons)
	Simple RTL instructions (less than a dozen registers)
	Parallel bus with data strobe and read/write lines, 5-bit register and 16-bit data lines
	I2C serial bus (100 and 400 Kbits)
	Digital input bus for use with the recognition stage built into the chip
	1 80-pin hirose connector for single or top-of-stack module
	Two 80-pin hirose connectors for in-the-stack modules
	Single source power supply (5 to 12 volts)
	Perform an equivalent 100 Gigaoperations per second @ 15 MHz
	Average power with power management 0.5 watt

Applications
		Object recognition Surface anomaly detection Biometrics Target tracking Video surveillance Smart motion detection Satellite imaging Medical imaging Speech recognition Radar identification EKG, EEG monitoring Sonar identification Spectrum recognition Vibration monitoring Data mining More....

		The CogniMem chip implements two powerful non-linear classifiers (RCE and KNN) in a natively parallel architecture. The tremendous benefit of this architecture is a recognition cycle which remains under 11 us whether the entire network is composed of 1, 2 or more chips. Brute computational power is equivalent to 80 gig operations/second @ 27 MHz for one chip, twice as many for two chips, etc. If the real-time recognition engine built into CogniMem is running, the data received on the digital input bus is automatically broadcasted to the neurons and the response of the neuron with the best match is available within less than 11 microseconds after the feed of the last data.

CM1K neural network Parallel architecture with 1024 neurons RCE (Restricted Coulomb Energy) Two classifiers: Radial Basis Function (RBF) K-Nearest Neighbor (KNN) classifier Vector data: up to 256 bytes Classification status: Identified (pin 13), Uncertain (pin 14) or Unknown Categories: up to 32768 values Distance calculation: L1 or LSup distance norms Sub-networks: up to 127 context values Trained by example Recognition stage for direct digital input Connectivity Parallel bus Serial bus I2C (100 kbit, 400 kbit and 3.3Mbit) Male and female 80-pin hirose connector for stackability Compatible with the CogniMem base boards Up to four stacked modules, totaling 8192 neurons Electrical / mechanical 300 mW @ 27 MHz Single source power supply (5 to 12 volts) Average power management w/ 0.5 watt High speed recognition stage for digital input bus V_clock up to 27 Mhz (pin 15) 8-bit data (pins 18-25) 1 sync. line for vector input (pin 16) 2 sync. Lines for video input (pins 16-17) 1 output strobe signal (pin 3) after each vector recognition Digital output bus (pins 4-10)		High speed recognition stage for digital input bus V_clock up to 27 Mhz (pin 15) 8-bit data (pins 18-25) 1 sync. line for vector input (pin 16) 2 sync. Lines for video input (pins 16-17) 1 output strobe signal (pin 3) after each vector recognition Digital output bus (pins 4-10) Timings @ 27 Mhz with default vectors of 256 bytes Learning time 10 µsec (275 cc) Recognition status in 8 µsec (257 cc) Best match in 11 µs (275 cc) Subsequent match in 3 µs/match (35 cc) High speed recognition stage for digital input bus V_clock up to 27 Mhz (pin 15) 8-bit data (pins 18-25) 1 sync. line for vector input (pin 16) 2 sync. Lines for video input (pins 16-17) 1 output strobe signal (pin 3) after each vector recognition Digital output bus (pins 4-10)

Purchase The	CM-EMB1K with 1024 neurons »
	CM-EMB2K, Top module with 2048 neurons »
	CM-EMB2K-I, 2048 neurons, in-the-stack »
For more information, you can download some documentations from HERE.

Please Click Here For Your Free C4AV Catalog

Image Sensors
Image Processors/SOC

CogniMem Technology

Center for Artificial Vision Product Selection Guide

Pattern Recognition Demo

Digital Camera & Pattern Recognition Seminars

Application Notes &
Meetings from Aptina

Research Material &
White Papers

C4AV will be joining
with a local Michigan University in the very
near future to provide course work & research
project support.

Sensory Intelligence Workshop at the University of Hawaii

Center for Global Sensory Intelligence Workshop

Privacy Statement - Terms of Use