Predicting Deep Zero-Shot Convolutional Neural Networks using Textual Descriptions

TL;DR

This paper introduces a model that predicts CNN weights from textual descriptions, enabling zero-shot classification of unseen categories using Wikipedia articles, and outperforms previous methods on bird and flower datasets.

Contribution

The novel approach predicts CNN layer weights from text features, leveraging CNN architecture for improved zero-shot learning from textual descriptions.

Findings

Significantly outperforms previous zero-shot learning methods.

Automatically generates pseudo-attributes from Wikipedia articles.

Effective on bird and flower datasets with ROC and Precision-Recall metrics.

Abstract

One of the main challenges in Zero-Shot Learning of visual categories is gathering semantic attributes to accompany images. Recent work has shown that learning from textual descriptions, such as Wikipedia articles, avoids the problem of having to explicitly define these attributes. We present a new model that can classify unseen categories from their textual description. Specifically, we use text features to predict the output weights of both the convolutional and the fully connected layers in a deep convolutional neural network (CNN). We take advantage of the architecture of CNNs and learn features at different layers, rather than just learning an embedding space for both modalities, as is common with existing approaches. The proposed model also allows us to automatically generate a list of pseudo- attributes for each visual category consisting of words from Wikipedia articles. We train our models end-to-end us- ing the Caltech-UCSD bird and flower datasets and evaluate both ROC and Precision-Recall curves. Our empirical results show that the proposed model significantly outperforms previous methods.