Simone Bonechi (DIISM, University of Siena)
May 10, 2018 – 9:30 AM
DIISM, Artificial Intelligence laboratory (room 201), Siena SI
Most of the leading convolutional neural networks for semantic segmentation exploit a large number of pixel–level annotations. Such human based labelling require a considerable effort that complicate the creation of large–scale datasets. In this seminar I will present a deep learning approach that uses bounding box annotations to obtain semantic segmentation. The proposed method is based on a two stage training procedure; in the first stage a deep neural network
is trained to distinguish the relevant object from the background inside bounding box, in the second stage the output of the first network is used to provide a weak supervision for a multi–class segmentation CNN. The performances of our approach has been assessed on the Pascal–VOC 2012 segmentation dataset, obtaining competitive results compared to a fully supervised setting.