Aligned Image-Word Representations Improve Inductive Transfer Across Vision-Language Tasks.
Tanmay GuptaKevin J. ShihSaurabh SinghDerek HoiemPublished in: ICCV (2017)
Keyphrases
- input image
- image features
- image data
- computer vision
- image content
- image classification
- image segmentation
- image representation
- keywords
- context sensitive
- multiple tasks
- inductive transfer
- learning algorithm
- multiple task learning
- image retrieval
- prior knowledge
- active learning
- natural language
- training data
- supervised learning
- text categorization