Paper
reports, processed data, and consulting : applied maths study made by
NEXYAD for you,
on the basis of your task book plus (usually) a
set of data : : Artificial vision,
detection, pattern recognition, image
processing, signal processing and understanding, virtual sensors, data
analysis, statistics, modeling, ... Click HERE to read examples. |
The
Ultimate
pattern recognition application generator : RecogNexTM
RecogNexTM is a professional software that AUTOMATICALLY generates effective pattern recognition applications. Inputs of this tool are rough data and preprocessed data (using YOUR descriptors to preprocess data). From this external input, RecogNex automatically computes new preprocessings (neural networks based automatic information extraction), and edits statistics about the discrimination power of rough data, of your preprocessed data, and of data preprocessed by the automatic RecogNexTM preprocessing module. Those input data are used together by a fuzzy sets-based pattern recognition system that is also automatically calibrated (using advanced statistics + stochastic learning). Several pattern recognition systems are automatically built (with all inputs, the minimum possible number of inputs, ...), and they are automatically tested and compared. In the end, RecogNexTM generates the C code of the best pattern |
recognition application.
All you need in order to apply RecogNexTM
is a data base
with your set of quantitative and qualitative variables among
individuals, plus a qualitative variable giving the class number. You
can also use other qualitative variables that should describe the
context (ex : 1 = it was raining, 2 = the weather was fine, ...), so
the learning and test data bases builder (this is also automatic) will
take into account quotas and statistical constraints for representative
sampling using this context description : your data bases will be as
most representative as possible => robustness is optimized. Software on request Special software made for you (structured upon our applied maths algorithms library). The developed software is entirely specialized in solving efficiently a given recurrent project : you can run it on your own data. Vocabulary of the user interface uses professional words of YOUR job : no need to hire a mathematician to use the software |
These
systems
include sensors, hardware, software - applied-maths models of
human vision behaviour, mechanics, ... We propose 3 integrated systems for the wiper systems tests and tuning : - RainNexTM: RainNex is a programmable rain machine (with artificial rain very similar to natural rain : no need to wait for the rain anymore). With RainNex, it is possible to have exactly the same statistics of rain on the driver's vision area and on the rain sensor area (whatever the location of the rain sensor is). VisiNexTM : VisiNex lets you evaluate among time distance of visibility and visual quality score through the windscreen. Wiper is supposed to recover visibility : this recovery is "measured" (quantitative and absolute marks of visibility that allow the comparison/benchmarking of several wiper systems |
(including rain sensors, decision making software, quality of rubbers, ... DropNexTM : DropNex lets you count among time rain drops on a windscreen (size, statistics, ...). You can specify a rain machine (measure rains outdoor, and then provide statistics of rain as a target for a future rain machine). You can sharply tune your rain machine in order to reproduce interesting kinds of rain. You can analyse cases of "wiper failure" and know exactly which kind of rain occurs in such a case. |
On-board
algorithms for Advanced Driver Assistance Systems NEXYAD developed 4 algorithms that are ready for integration into your on-board applications : we can sell the technology (licence for R&D use, licence for integration into your products, technology transfert, property transfert, ...). |
Most of these algorithms are
available in
two versions : stand alone dll, or ".pck" components for the real time
simulation and development environment RT-MAPS
(from the French company Intempora).
Three algorithms are artificial vision ones : - obstacle detection (from a single camera) - road detection (from a single camera) - visibility estimation (from a single camera) |
Real
time image binarization algorithm for smart cameras NEXYAD developed a very fast binarization algorithm (very low computing and memory resources requested) for large image sequences that leads to 1-bit per pixel data flow and keeps the scene understandable for human eye. Perfect for real time remote monitoring applications and decision making from smart cameras. |