NAS-SpatialFlow : neural architecture search for panoptic segmentation

Neural architecture search (NAS) has achieved success in various deep learning tasks. NAS can automatically find an efficient neural network architecture for a certain task on a dataset, which can outperform human-designed neural architectures. NAS has proved its efficiency in classification task (I...

Full description

Saved in:
Bibliographic Details
Main Author: Cao, Liu
Other Authors: Chen Change Loy
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2020
Subjects:
Online Access:https://hdl.handle.net/10356/138013
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-138013
record_format dspace
spelling sg-ntu-dr.10356-1380132020-04-22T01:39:40Z NAS-SpatialFlow : neural architecture search for panoptic segmentation Cao, Liu Chen Change Loy School of Computer Science and Engineering SenseTime ccloy@ntu.edu.sg Engineering::Computer science and engineering Neural architecture search (NAS) has achieved success in various deep learning tasks. NAS can automatically find an efficient neural network architecture for a certain task on a dataset, which can outperform human-designed neural architectures. NAS has proved its efficiency in classification task (ImageNet). However, no previous work re- lated to NAS has been done on panoptic segmentation, a challenging task which unifies instance segmentation and semantic segmentation. In this final year project, we apply a gradient-based NAS method called DARTS to search for a network structure to bet- ter fuse features between sub-networks based on SpatiaFlow, a unified neural network with four sub-networks for panoptic segmentation. We name the final searched model NAS-SpatialFlow. Experimental results on the MS-COCO dataset show that NAS- SpatialFlow achieves 39.7 PQ, which is inferior to the human-designed model Spa- tialFlow (40.5 PQ). Our results demonstrate the ineffectiveness of the current gradient- based NAS methods in more complicated tasks like panoptic segmentation. The main contributions of this final year project are three-fold (1) We reimplement SpatialFlow as our baseline model and prove its effectiveness on panoptic segmenta- tion. (2) We apply NAS methods to search for network structures for feature fusion between sub-networks (3) We design attention modules to further enhance the recog- nition of the background stuff of SpatialFlow. Bachelor of Engineering (Computer Science) 2020-04-22T01:39:40Z 2020-04-22T01:39:40Z 2020 Final Year Project (FYP) https://hdl.handle.net/10356/138013 en SCSE19-0112 application/pdf Nanyang Technological University
institution Nanyang Technological University
building NTU Library
country Singapore
collection DR-NTU
language English
topic Engineering::Computer science and engineering
spellingShingle Engineering::Computer science and engineering
Cao, Liu
NAS-SpatialFlow : neural architecture search for panoptic segmentation
description Neural architecture search (NAS) has achieved success in various deep learning tasks. NAS can automatically find an efficient neural network architecture for a certain task on a dataset, which can outperform human-designed neural architectures. NAS has proved its efficiency in classification task (ImageNet). However, no previous work re- lated to NAS has been done on panoptic segmentation, a challenging task which unifies instance segmentation and semantic segmentation. In this final year project, we apply a gradient-based NAS method called DARTS to search for a network structure to bet- ter fuse features between sub-networks based on SpatiaFlow, a unified neural network with four sub-networks for panoptic segmentation. We name the final searched model NAS-SpatialFlow. Experimental results on the MS-COCO dataset show that NAS- SpatialFlow achieves 39.7 PQ, which is inferior to the human-designed model Spa- tialFlow (40.5 PQ). Our results demonstrate the ineffectiveness of the current gradient- based NAS methods in more complicated tasks like panoptic segmentation. The main contributions of this final year project are three-fold (1) We reimplement SpatialFlow as our baseline model and prove its effectiveness on panoptic segmenta- tion. (2) We apply NAS methods to search for network structures for feature fusion between sub-networks (3) We design attention modules to further enhance the recog- nition of the background stuff of SpatialFlow.
author2 Chen Change Loy
author_facet Chen Change Loy
Cao, Liu
format Final Year Project
author Cao, Liu
author_sort Cao, Liu
title NAS-SpatialFlow : neural architecture search for panoptic segmentation
title_short NAS-SpatialFlow : neural architecture search for panoptic segmentation
title_full NAS-SpatialFlow : neural architecture search for panoptic segmentation
title_fullStr NAS-SpatialFlow : neural architecture search for panoptic segmentation
title_full_unstemmed NAS-SpatialFlow : neural architecture search for panoptic segmentation
title_sort nas-spatialflow : neural architecture search for panoptic segmentation
publisher Nanyang Technological University
publishDate 2020
url https://hdl.handle.net/10356/138013
_version_ 1681058100171243520