Zero-shot semantic segmentation using relation network
Authors
Date
2020Copyright
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Zero-shot learning (ZSL) is widely studied in recent years to solve the problem of lacking annotations. Currently, most studies on ZSL are for image classification and object detection. But zero-shot semantic segmentation, pixel level classification, is still at its early stage. Therefore, this thesis proposes to extend a zero-shot image classification model, Relation Network (RN), to semantic segmentation tasks. This thesis modifies the structure of RN based on other state-of-the-arts semantic segmentation models (i.e. U-Net and DeepLab) and utilizes word embeddings from Caltech-UCSD Birds 200-2011 attributes and natural language processing models (i.e. word2vec and fastText). Because meta-learning is limited to binary tasks, this thesis proposes to join multiple binary semantic segmentation pipelines for multi-class semantic segmentation. It is proved by experiments that RN could improve accuracy of U-Net with the help of semantic side information on binary semantic segmentation and it could also be applied on multi-class semantic segmentation with simpler structure than the baseline model, SPNet, but higher accuracy under ZSL setting. However, the capability of RN under generalized zero-shot learning (GZSL) setting still needs improvement. This thesis also studies on how different word embeddings, network structures and data affect RN and what could be done to improve its results.
...


Keywords
Metadata
Show full item recordCollections
- Pro gradu -tutkielmat [24987]
Related items
Showing items with similar title or keywords.
-
Editorial for the special issue "Frontiers in spectral imaging and 3D technologies for geospatial solutions"
Honkavaara, Eija; Karantzalos, Konstantinos; Liang, Xinlian; Nocerino, Erica; Pölönen, Ilkka; Rönnholm, Petri (MDPI, 2019)This Special Issue hosts papers on the integrated use of spectral imaging and 3D technologies in remote sensing, including novel sensors, evolving machine learning technologies for data analysis, and the utilization of ... -
Zero-shot Semantic Segmentation using Relation Network
Zhang, Yindong; Khriyenko, Oleksiy (FRUCT Oy, 2021)Zero-shot learning (ZSL) is widely studied in recent years to solve the problem of lacking annotations. Currently, most studies on ZSL are for image classification and object detection. But, zero-shot semantic segmentation, ... -
Deep semantic segmentation for skin cancer detection from hyperspectral images
Karhu, Anette (2020)As skin cancer types are a growing concern worldwide, a new screening tool combined with automation may help the clinicians in clinical examinations of lesions. A novel hyperspectral imager prototype has been noted to be ... -
Multilabel segmentation of cancer cell culture on vascular structures with deep neural networks
Rahkonen, Samuli; Koskinen, Emilia; Pölönen, Ilkka; Heinonen, Tuula; Ylikomi, Timo; Äyrämö, Sami; Eskelinen, Matti A. (SPIE, 2020)New increasingly complex in vitro cancer cell models are being developed. These new models seem to represent the cell behavior in vivo more accurately and have better physiological relevance than prior models. An efficient ... -
Towards a Great Design of Conceptual Modelling
Kiyoki, Yasushi; Thalheim, Bernhard; Duží, Marie; Jaakkola, Hannu; Chawakitchareon, Petchporn; Heimbürger, Anneli (IOS Press, 2020)Humankind faces a most crucial mission; we must endeavour, on a global scale, to restore and improve our natural and social environments. This is a big challenge for global information systems development and for their ...