ICCV 2015 RIDE: Reversal invariant Descriptor Enhancement Speaker: Lingxi Xie Authors: Lingxi Xie, Jingdong Wang Weiyao Lin, Bo Zhang, Qi Tian State Key Laboratory of Intelligent Technology and Systems Department of Computer Science and Technology Tsinghua University http://www.tsinghua.edu.cn
ICCV 2015 RIDE: Reversal Invariant Descriptor Enhancement Speaker: Lingxi Xie Authors: Lingxi Xie, Jingdong Wang, Weiyao Lin, Bo Zhang, Qi Tian State Key Laboratory of Intelligent Technology and Systems Department of Computer Science and Technology Tsinghua University http://www.tsinghua.edu.cn
Outline ntroduction The bag- of-Feature model The ride algorithm Motivation Towards reversal invariance Experimental Results · Conclusions 1/26/2021 ICCV 2015-Presentation
Outline • Introduction • The Bag-of-Feature Model • The RIDE Algorithm – Motivation – Towards Reversal Invariance • Experimental Results • Conclusions 1/26/2021 ICCV 2015 - Presentation 2
Outline ntroduction The bag-of-Feature Model The ride algorithm Motivation Towards reversal invariance Experimental Results · Conclusions 1/26/2021 ICCV 2015-Presentation
Outline • Introduction • The Bag-of-Feature Model • The RIDE Algorithm – Motivation – Towards Reversal Invariance • Experimental Results • Conclusions 1/26/2021 ICCV 2015 - Presentation 3
Image classification 1/26/2021 ICCV 2015-Presentation
Image Classification 1/26/2021 ICCV 2015 - Presentation 4
Outline ntroduction The Bag-of-Feature Model The ride algorithm Motivation Towards reversal invariance Experimental Results · Conclusions 1/26/2021 ICCV 2015-Presentation
Outline • Introduction • The Bag-of-Feature Model • The RIDE Algorithm – Motivation – Towards Reversal Invariance • Experimental Results • Conclusions 1/26/2021 ICCV 2015 - Presentation 5
Image-level Vector Spatial Pooling Sum Pooling/Max Pooling Spatial Pyramid Matching [Lazebnik, CVPRO6 Compact Geometric Phrase Pooling Xie, ACMMM12 Feature codes Hard/Soft/Sparse Coding methods Vector Quantization LLC Encoding [Wang, CVpR10 Visual Fisher Vector Encoding [perronnin ECCv10 Vocabula Clustering Methods K-M Hierarchical K-Means [Nister, CVPRO6 Image Approximate K-Means[Philbin, CVPRO7 Descriptors Gradient-based Local descriptors SIFT [Lowe, IJCV04 HOG [DalaL, CVPRO5 LCS[Perronnin, ECCV10 Raw Image 1/16/2021 ICCV 2015-Presentation
1/26/2021 ICCV 2015 - Presentation 6 Raw Image Image Descriptors Visual Vocabulary Compact Feature Codes Image-level Vector Gradient-based Local Descriptors: SIFT [Lowe, IJCV04] HOG [Dalal, CVPR05] LCS [Perronnin, ECCV10] Clustering Methods: K-Means Hierarchical K-Means [Nister, CVPR06] Approximate K-Means [Philbin, CVPR07] Hard/Soft/Sparse Coding methods: Vector Quantization LLC Encoding [Wang, CVPR10] Fisher Vector Encoding [Perronnin, ECCV10] Spatial Pooling: Sum Pooling/Max Pooling, Spatial Pyramid Matching [Lazebnik, CVPR06] Geometric Phrase Pooling [Xie, ACMMM12]
Image-level Vector Spatial Pooling Sum Pooling/Max Pooling Spatial Pyramid Matching [Lazebnik, CVPRo6 Compact Geometric Phrase Pooling Xie, ACMMM12 Feature codes Hard/Soft/Sparse Coding methods Vector Quantization LLC Encoding [Wang, CVPR10 Visual Fisher Vector Encoding [Perronnin, ECCV10 Vocabulary Clustering Methods K-Means Hierarchical K-Means [Nister, CVPRO6 Image Approximate K-Means[Philbin, CVPRO7 Descriptors Gradient-based Local Descriptors SIFT [Lowe, IJCV04 HOG [DalaL, CVPRO5 LCS [Perronnin, ECCV10] Raw Image 1/16/2021 ICCV 2015-Presentation
1/26/2021 ICCV 2015 - Presentation 7 Raw Image Image Descriptors Visual Vocabulary Compact Feature Codes Image-level Vector Gradient-based Local Descriptors: SIFT [Lowe, IJCV04] HOG [Dalal, CVPR05] LCS [Perronnin, ECCV10] Clustering Methods: K-Means Hierarchical K-Means [Nister, CVPR06] Approximate K-Means [Philbin, CVPR07] Hard/Soft/Sparse Coding methods: Vector Quantization LLC Encoding [Wang, CVPR10] Fisher Vector Encoding [Perronnin, ECCV10] Spatial Pooling: Sum Pooling/Max Pooling, Spatial Pyramid Matching [Lazebnik, CVPR06] Geometric Phrase Pooling [Xie, ACMMM12]
Outline ntroduction The bag- of-Feature model The ride algorithm Motivation Towards reversal invariance Experimental Results · Conclusions 1/26/2021 ICCV 2015-Presentation 8
Outline • Introduction • The Bag-of-Feature Model • The RIDE Algorithm – Motivation – Towards Reversal Invariance • Experimental Results • Conclusions 1/26/2021 ICCV 2015 - Presentation 8
Image Matching Reversal Copy What We Want What SiFT Does 1/26/2021 ICCV 2015-Presentation
Image Matching: Reversal Copy 1/26/2021 ICCV 2015 - Presentation 9 What We Want What SIFT Does
Image Matching Reversal objects What We Want What SiFT Does 1/26/2021 ICCV 2015-Presentation
Image Matching: Reversal Objects 1/26/2021 ICCV 2015 - Presentation 10 What We Want What SIFT Does