Dual-path Convolutional Image-Text Embeddings with Instance Loss

Autor: Liang Zheng, Yi-Dong Shen, Zhedong Zheng, Yi Yang, Mingliang Xu, Michael Garrett
Rok vydání: 2020
Předmět:
FOS: Computer and information sciences
Computer Networks and Communications
Computer science
Computer Vision and Pattern Recognition (cs.CV)
Computer Science - Computer Vision and Pattern Recognition
Initialization
02 engineering and technology
Machine learning
computer.software_genre
Convolutional neural network
Field (computer science)
Ranking (information retrieval)
Discriminative model
0803 Computer Software
0805 Distributed Computing
0806 Information Systems

Margin (machine learning)
0202 electrical engineering
electronic engineering
information engineering

Artificial Intelligence & Image Processing
Word2vec
business.industry
020207 software engineering
Multimedia (cs.MM)
Hardware and Architecture
Feature (computer vision)
020201 artificial intelligence & image processing
Artificial intelligence
business
computer
Computer Science - Multimedia
Zdroj: ACM Transactions on Multimedia Computing, Communications, and Applications. 16:1-23
ISSN: 1551-6865
1551-6857
DOI: 10.1145/3383184
Popis: Matching images and sentences demands a fine understanding of both modalities. In this paper, we propose a new system to discriminatively embed the image and text to a shared visual-textual space. In this field, most existing works apply the ranking loss to pull the positive image / text pairs close and push the negative pairs apart from each other. However, directly deploying the ranking loss is hard for network learning, since it starts from the two heterogeneous features to build inter-modal relationship. To address this problem, we propose the instance loss which explicitly considers the intra-modal data distribution. It is based on an unsupervised assumption that each image / text group can be viewed as a class. So the network can learn the fine granularity from every image/text group. The experiment shows that the instance loss offers better weight initialization for the ranking loss, so that more discriminative embeddings can be learned. Besides, existing works usually apply the off-the-shelf features, i.e., word2vec and fixed visual feature. So in a minor contribution, this paper constructs an end-to-end dual-path convolutional network to learn the image and text representations. End-to-end learning allows the system to directly learn from the data and fully utilize the supervision. On two generic retrieval datasets (Flickr30k and MSCOCO), experiments demonstrate that our method yields competitive accuracy compared to state-of-the-art methods. Moreover, in language based person retrieval, we improve the state of the art by a large margin. The code has been made publicly available.
Comment: 15pages, 15 figures, 8 tables
Databáze: OpenAIRE