Learning Visual Features from Large Weakly Supervised Data
This work addresses the bottleneck of data labeling in computer vision, offering a scalable alternative to supervised learning, though it is incremental as it builds on existing weakly supervised methods.
The paper tackled the problem of limited manually labeled data for training convolutional networks by leveraging a massive, weakly-labeled dataset of 100 million Flickr photos and captions, resulting in features that perform well across various vision tasks and capture semantic relationships like word similarity and cross-language correspondences.
Convolutional networks trained on large supervised dataset produce visual features which form the basis for the state-of-the-art in many computer-vision problems. Further improvements of these visual features will likely require even larger manually labeled data sets, which severely limits the pace at which progress can be made. In this paper, we explore the potential of leveraging massive, weakly-labeled image collections for learning good visual features. We train convolutional networks on a dataset of 100 million Flickr photos and captions, and show that these networks produce features that perform well in a range of vision problems. We also show that the networks appropriately capture word similarity, and learn correspondences between different languages.