CONTEXT
HIPE is a modular framework presenting the entire signal processing by a visual and graphic workflow system. From the start, we wanted Hipe to be an Open source asset in order to build a community outside Aldwin.
The idea came out after observing the constant effort made by developers in that field of work : every time, they had to the exact same work over and over again. That’s when we decided to create HIPE. HIPE helps you to increase the performance of your developments by increasing your delivery time in response to a customer’s needs.
OBJECTIF
At first, the main goal was to leverage the know-how in the field of image and real-time video processing in order to re-share it later on.
To be able to build on this know-how offers various benefits :
– Avoid loss of previous work
– Reuse previous work
– Reuse this know-how involving different parameters and calibration but applied to similar business cases (time saving).
– Make accessible to high performance material resources such as IBM Power 9 to those who do not have the skills in their organisation.
– Offer the opportunity to embed proprietary developments while using external library (i.e. any image processing library or other A.I. analysis tools)
– Leverage specialized know-how in different sectors such as industry, hospital or security and to be able to reuse it with various clients.
– Increase the processing scalability and ability port them to grid computing or clou
– The majority of module at the heart of Hipe are under an AGPL license, however this framework gives you the opportunity to develop proprietary code with copyright that cannot be disclosed to the public or used by Aldwin.
RESULTS
We have worked on various use cases:
– Poclain: Defect detection on hydraulic motor
– Biometrics to recognize emotions and facial positioning
– Detection of marked cancer cells in organic tissues in partnership with the University of Paris Descartes and a Histology Department
– Porting of an existing bacterial phenotype recognition solution in order to industrialize it and classify 98% of bacteria.
Our trusting partners : IBM, NVIDIA, INGEDATA, INTEL, AWS, GOOGLE, L’université Paris Descartes
For more info : https://www.aneo.eu/hipe/
