clEsperanto : A unified architecture for GPU-accelerated image processing library
Stéphane Rigaud, Image Analysis Hub, C2RI, Institut Pasteur, Paris, France; Robert Haase, DFG Cluster of Excellence Physics of Life, TU Dresden, Germany
I2K 2022 | Talk Session #2 | May 6th
clEsperanto project, still in prototype stage, is the next-level incarnation of CLIJ. It aims at making GPU-accelerated image processing available to all the major languages and software used in Bio-Image Analysis, from Fiji, Icy, MATLAB and Java to Napari and Python, with a single common core. This library relies on an architecture of language specific top-layer API wrapped around CLIc, a generic low-level C++ layer in charge of managing processing operations and device-host communications. Altogether, it improve re-usability, reproducibility, and maintainability, while drastically lowering the learning curve for using GPU-acceleration in Bio-Image Analysis. In this talk we will present clEsperanto, discuss the library new architecture and performance, and showcase a few of its key capabilities on biological data.
Keywords: Image Processing, GPU acceleration