Learning to Measure Change: Fully Convolutional Siamese Metric Networks for Scene Change Detection

Guo, Enqiang; Fu, Xinsha; Zhu, Jiawei; Deng, Min; Liu, Yu; Zhu, Qing; Li, Haifeng

doi:10.48550/arxiv.1810.09111

Cited by 22 publications

(40 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Backbone F1-SCORE TSUNAMI GSV AVERAGE EFnet [1] U-Net 0.659 0.565 0.612 Siam-Conc [1] U-Net 0.709 0.638 0.673 Siam-Diff [1] U-Net 0.717 0.647 0.682 CosimNet [24] DeepLabV2 0.806 0.692 0.749 CDNet [4] U-Net 0.838 0.693 0.766 CDNet++ [10] VGG-19 0.860 0.680 0.770 CSCDNet [5] ResNet-18 0.859 0.738 0.799 DR-TANet (Ours)…”

Section: Networkmentioning

confidence: 99%

See 1 more Smart Citation

DR-TANet: Dynamic Receptive Temporal Attention Network for Street Scene Change Detection

Chen¹,

Yang²,

Stiefelhagen³

2021

Preprint

View full text Add to dashboard Cite

Street scene change detection continues to capture researchers' interests in the computer vision community. It aims to identify the changed regions of the paired street-view images captured at different times. The state-of-the-art network based on the encoder-decoder architecture leverages the feature maps at the corresponding level between two channels to gain sufficient information of changes. Still, the efficiency of feature extraction, feature correlation calculation, even the whole network requires further improvement. This paper proposes the temporal attention and explores the impact of the dependencyscope size of temporal attention on the performance of change detection. In addition, based on the Temporal Attention Module (TAM), we introduce a more efficient and light-weight version -Dynamic Receptive Temporal Attention Module (DRTAM)and propose the Concurrent Horizontal and Vertical Attention (CHVA) to improve the accuracy of the network on specific challenging entities. On street scene datasets 'GSV', 'TSUNAMI' and 'VL-CMU-CD', our approach gains excellent performance, establishing new state-of-the-art scores without bells and whistles, while maintaining high efficiency applicable in autonomous vehicles.

show abstract

Section: Networkmentioning

confidence: 99%

“…Backbone F1-SCORE EFnet [1] U-Net 0.581 Siam-Conc [1] U-Net 0.664 Siam-Diff [1] U-Net 0.625 CosimNet [24] DeepLabV2 0.706 CDNet [4] U-Net 0.685 CSCDNet [5] ResNet-18 0.710 DR-TANet (Ours)…”

Section: Networkmentioning

confidence: 99%

DR-TANet: Dynamic Receptive Temporal Attention Network for Street Scene Change Detection

Chen¹,

Yang²,

Stiefelhagen³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Siamese networks have been used in various applications of indoor design and floorplanning due to its ability to learn from limited data. For example, [12] used Siamese net-works for scene change detection.…”

Section: Siamese Networkmentioning

confidence: 99%

Contextual Scene Augmentation and Synthesis via GSACNet

Keshavarzi¹,

Christian²,

Shrivastava³

et al. 2021

Preprint

View full text Add to dashboard Cite

Indoor scene augmentation has become an emerging topic in the field of computer vision and graphics with applications in augmented and virtual reality. However, current state-of-the-art systems using deep neural networks require large datasets for training. In this paper we introduce GSACNet, a contextual scene augmentation system that can be trained with limited scene priors. GSACNet utilizes a novel parametric data augmentation method combined with a Graph Attention and Siamese network architecture followed by an Autoencoder network to facilitate training with small datasets. We show the effectiveness of our proposed system by conducting ablation and comparative studies with alternative systems on the Matterport3D dataset. Our results indicate that our scene augmentation outperforms prior art in scene synthesis with limited scene priors available.

show abstract

“…Change Detection using Deep Learning. More recently, several works [1], [15]- [19] apply deep learning to detect scene changes. These works learn to compare two input images to detect changes in a strongly supervised manner, requiring pixel level [1], [15], [18], [19] or patch level [17] annotations which can be highly tedious to obtain.…”

Section: Related Workmentioning

confidence: 99%

“…More recently, several works [1], [15]- [19] apply deep learning to detect scene changes. These works learn to compare two input images to detect changes in a strongly supervised manner, requiring pixel level [1], [15], [18], [19] or patch level [17] annotations which can be highly tedious to obtain. To avoid the need for annotation, Sakurada and Okatani [20] compare normalized features extracted from an upper layer of a convolutional neural network pretrained on a image recognition task, and make use of super-pixel segmentations to obtain high resolution outputs.…”

Section: Related Workmentioning

confidence: 99%

City-scale Scene Change Detection using Point Clouds

Yew

Lee

2021

2021 IEEE International Conference on Robotics and Automation (ICRA)

View full text Add to dashboard Cite

We propose a method for detecting structural changes in a city using images captured from vehicular mounted cameras over traversals at two different times. We first generate 3D point clouds for each traversal from the images and approximate GNSS/INS readings using Structure-from-Motion (SfM). A direct comparison of the two point clouds for change detection is not ideal due to inaccurate geo-location information and possible drifts in the SfM. To circumvent this problem, we propose a deep learning-based non-rigid registration on the point clouds which allows us to compare the point clouds for structural change detection in the scene. Furthermore, we introduce a dual thresholding check and post-processing step to enhance the robustness of our method. We collect two datasets for the evaluation of our approach. Experiments show that our method is able to detect scene changes effectively, even in the presence of viewpoint and illumination differences.

show abstract

Learning to Measure Change: Fully Convolutional Siamese Metric Networks for Scene Change Detection

Cited by 22 publications

References 37 publications

DR-TANet: Dynamic Receptive Temporal Attention Network for Street Scene Change Detection

DR-TANet: Dynamic Receptive Temporal Attention Network for Street Scene Change Detection

Contextual Scene Augmentation and Synthesis via GSACNet

City-scale Scene Change Detection using Point Clouds

Contact Info

Product

Resources

About