Zero-Shot Semantic Segmentation

Abstract : Semantic segmentation models are limited in their ability to scale to large numbers of object classes. In this paper, we introduce the new task of zero-shot semantic segmentation: learning pixel-wise classifiers for never-seen object categories with zero training examples. To this end, we present a novel architecture, ZS3Net, combining a deep visual segmentation model with an approach to generate visual representations from semantic word embeddings. By this way, ZS3Net addresses pixel classification tasks where both seen and unseen categories are faced at test time (so called "generalized" zero-shot classification). Performance is further improved by a self-training step that relies on automatic pseudo-labeling of pixels from unseen classes. On the two standard segmentation datasets, Pascal-VOC and Pascal-Context, we propose zero-shot benchmarks and set competitive baselines. For complex scenes as ones in the Pascal-Context dataset, we extend our approach by using a graph-context encoding to fully leverage spatial context priors coming from class-wise segmentation maps.
Document type :
Preprints, Working Papers, ...
Complete list of metadatas

Cited literature [49 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-02146433
Contributor : Maxime Bucher <>
Submitted on : Monday, June 3, 2019 - 11:41:43 PM
Last modification on : Friday, July 5, 2019 - 3:26:03 PM

File

Zero_Shot_Semantic_Segmentatio...
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02146433, version 1

Citation

Maxime Bucher, Tuan-Hung Vu, Matthieu Cord, Patrick Pérez. Zero-Shot Semantic Segmentation. 2019. ⟨hal-02146433⟩

Share

Metrics

Record views

20

Files downloads

167