Webly Supervised Learning of Convolutional Networks

Successful semantic segmentation methods typically rely on the training datasets containing a large number of pixel-wise labeled images. To alleviate the dependence on such a fully annotated training dataset, in this paper, we propose a semi- and weakly-...

In this paper, we propose a weakly supervised semantic segmentation method by directly learning from web images, which are crawled from the Internet by using text queries, without any explicit user annotation or even data filtering. With the goal ...

• We propose a weakly supervised semantic segmentation method by directly learning from web images without relying on any explicit user annotations or even data pre-selection.
• We propose a novel data complexity measurement for the webly-...

Deep learning usually requires large amounts of labeled training data, but annotating data is costly and tedious. The framework of semi-supervised learning provides the means to use both labeled data and arbitrary amounts of unlabeled data for ...