Deep Learning with 20 Images
Adaptive Vision has released a new version (4.10) of its deep learning flagship product, Adaptive Vision Studio, extended with a second generation of deep learning tools. It comes with significantly better performance as well as with a completely new tool for simultaneous object localization and classification.
‚Instance Segmentation‘ not only calculates object locations and class labels, but also precise regions corresponding to the objects. The new tool open new possibilities in application such us object counting (e.g. fruits) and robot picking (e.g. handling and packaging flowers). It is trained with as few as 20-50 sample images.