Deep Compositional Captioning: Describing Novel Object Categories without Paired Training Data

By Lisa Anne Hendricks, Subhashini Venugopalan, Marcus Rohrbach, Raymond J. Mooney, Kate Saenko, Trevor Darrell, Uc Berkeley. 2016

Subfields: Computer Vision and Pattern Recognition, Natural Language Generation

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Trevor Darrell
2. Luc Van Gool
3. Yi Yang
4. Dacheng Tao
5. Bernt Schiele
6. Xiaogang Wang
7. Andrew Zisserman
8. Ming-Hsuan Yang

In Natural Language Generation

1. Lisa Anne Hendricks
2. Kate Saenko
3. Rita Cucchiara
4. Lorenzo Baraldi
5. Marcella Cornia
6. Trevor Darrell
7. Giuseppe Fiameni
8. Marcus Rohrbach

Citations in the corpus

• Movie Description (2016, methodology): “We expect new techniques, including relying on different modalities, see e.g. (Hendricks et al. 2016), to overcome this challenge.”
• Object Hallucination in Image Captioning (2018, methodology): “In addition to the Robust Test set, we also consider a set of MSCOCO in which certain objects are held out, which we call the Novel Object split (Hendricks et al., 2016).”
• From Show to Tell: A Survey on Deep Learning-Based Image Captioning (2021, background): “To explore this strategy, Hendricks et al. [176] introduced a variant of the COCO dataset [128], called held-out COCO , in which image-caption pairs containing one of eight pre-selected object classes were removed from the training set but not from the test set.”
• From Show to Tell: A Survey on Image Captioning (2021, background): “To explore this strategy, Hendricks et al. [151] introduced a variant of the MS COCO dataset [108], called held-out COCO , in which image-caption pairs containing one of eight pre-selected object classes were removed from the training set but not from the test set.”
• Image Captioning with Very Scarce Supervised Data: Adversarial Semi-Supervised Learning Approach (2019, background): “Most of aforementioned works [Gu et al. , 2018; Anne Hendricks et al. , 2016; Venugopalan et al. , 2017; Feng et al. , 2019] exploit large auxiliary supervised data which is often beyond image-caption data.”
• Designing a Moral Compass for the Future of Computer Vision Using Speculative Analysis (2017, unknown): “Computer vision (CV) techniques are at the epicenter of excitement and progress related to recent developments in deep learning; specifically, convolutional neural networks [36, 29, 30, 64].”