The ability to rapidly updatemachine vision based inspection systems is paramount . We develop amethodology for learning actively, from rapidly mined, weakly (i.e. partially)annotated data, enabling a fast, direct feedback from the operators on theproduction line . We also consider the problem of covariate shift, which arises inevitably due tochanging conditions during data acquisition . In that regard, we show domain-adversarial training to be an efficient way to address this issue . We show domain adversarial training is efficient way of addressing this issue, and tackling a big machine vision weakness: false positives. In this work, we develop a new tool for machine vision systems to tackle false positives .

Author(s) : Antoine Cordier, Deepan Das, Pierre Gutierrez

Links : PDF - Abstract

Code :

Keywords : vision - false - learning - machine - data -

Leave a Reply

Your email address will not be published. Required fields are marked *