Foreground-Aware Relation Network for Geospatial Object Segmentation in High Spatial Resolution Remote Sensing Imagery

Zheng, Zhuo; Zhong, Yanfei; Wang, Junjue; Ma, Ailong

doi:10.1109/cvpr42600.2020.00415

Cited by 218 publications

(127 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Semantic segmentation is an important computer vision task that has advanced significantly due to deep learning techniques [1][2][3][4][5][6]. Most semantic segmentation methods focus on standard datasets, such as MS-COCO [7] and Cityscapes [8], but there is great potential in diverse applications such as remote sensing [9], agriculture [10] and food recognition [11,12]. Unfortunately, methods for food segmentation are still lagging in development and this paper aims to advance the state-of-the-art.…”

Section: Introductionmentioning

confidence: 99%

GourmetNet: Food Segmentation Using Multi-Scale Waterfall Features with Spatial and Channel Attention

Sharma

Artacho

Savakis

2021

Sensors

View full text Add to dashboard Cite

We propose GourmetNet, a single-pass, end-to-end trainable network for food segmentation that achieves state-of-the-art performance. Food segmentation is an important problem as the first step for nutrition monitoring, food volume and calorie estimation. Our novel architecture incorporates both channel attention and spatial attention information in an expanded multi-scale feature representation using our advanced Waterfall Atrous Spatial Pooling module. GourmetNet refines the feature extraction process by merging features from multiple levels of the backbone through the two attention modules. The refined features are processed with the advanced multi-scale waterfall module that combines the benefits of cascade filtering and pyramid representations without requiring a separate decoder or post-processing. Our experiments on two food datasets show that GourmetNet significantly outperforms existing current state-of-the-art methods.

show abstract

Section: Introductionmentioning

confidence: 99%

GourmetNet: Food Segmentation Using Multi-Scale Waterfall Features with Spatial and Channel Attention

Sharma

Artacho

Savakis

2021

Sensors

View full text Add to dashboard Cite

show abstract

“…Several state-of-the-art methods were selected as benchmarks, which include three U-shape-based variants, i.e., FC-EF-Res [28], Peng et al [29], and W-Net [34] and four attention-based methods, i.e., DANet [58], FarSeg [57], STANet [6], DDCNN [37]. In particular, STANet was proposed by creators of the LEVIR-CD dataset.…”

Section: ) Comparisons On Levir-cdmentioning

confidence: 99%

“…For comparison, four attention-based methods, i.e., DANet [58], FarSeg [57], STANet [6], DDCNN [37] and two U-shapebased variants, i.e., FC-EF-Res [28] and Peng et al [29], as well as the post-classification-based method proposed by Ji et al [11] were selected as benchmarks. Ji et al [11] are exactly the creators of the WHU dataset.…”

Section: ) Comparisons On Whumentioning

confidence: 99%

AGCDetNet:An Attention-Guided Network for Building Change Detection in High-Resolution Remote Sensing Images

Song

Jiang

2021

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

While deep learning-based methods have gained considerable improvements in remote sensing (RS) image change detection (CD), scale variations and pseudo-changes hinder most supervised methods' performance. The CD networks derived from other fields can be fronted with false alarms and miss detections in high-resolution RS images due to the weak feature representation ability. In this paper, an attention-guided end-to-end change detection network (AGCDetNet) is proposed based on the fully convolutional network and attention mechanism. AGCDetNet learns to enhance the feature representation of change information and achieve accuracy improvements using spatial-and channelattention. A spatial attention module (SPAM) promotes the discrimination between the changed objects and the background by adding the learned spatial attention to the deep features. Channel-wise attention-guided interference filtering unit (CIFU) / atrous spatial pyramid pooling module (CG-ASPP) enhances the representation of multilevel features and multiscale context, respectively. Extensive experiments have been conducted on several public datasets for performance evaluation, including LEVIR-CD, WHU, Season-Varying, WV2, and ZY3. Experiment results demonstrate that AGCDetNet outperforms several stateof-the-art methods of accuracy and robustness. Specifically, AGCDetNet achieves the best F1-score on two datasets, i.e., LEVIR-CD (0.9076) and Season-Varying (0.9654).

show abstract

“…2) Comparisons on LEVIR-CD. W-Net [3], FC-EF-Res [4], and Peng et al [5], and attention-based methods STANet [6], DDCNN [7], and FarSeg [17] were selected as benchmarks. In particular, STANet was proposed by the dataset's creator.…”

Section: Comparisons With Other Approaches 1) Comparisons Of Netwomentioning

confidence: 99%

3M-CDNet: A Universal Powerful Benchmark for Remote Sensing Image Change Detection

Song¹,

Jiang²

2021

Preprint

View full text Add to dashboard Cite

Abstract—While deep learning-based methods have gained popularity and have made remarkable progress in remote sensing (RS) image change detection (CD), the limited amount of available data hinders the performance of most supervised methods. The CD networks transferred or derived from other fields can be fronted with a weak generalization capability. Developing a universal benchmark for performance evaluations based on the available datasets is urgent. To address these problems, we proposed a lightweight network, termed 3M-CDNet, which only requires about 3.12 M parameters. The lighter the network, the easier it is to train and alleviate overfitting the limited amount of data, resulting in a better generalization capability. 3M-CDNet has a flexible modular design that achieves performance improvements by incorporating plug-and-play modules. 3M-CDNet gains accuracy improvements in two ways: (1) the application of deformable convolutions (DConv) in the backbone network to gain a good geometric transformation modeling capacity for CD and (2) the application of an effective two-level feature fusion strategy to enhance the feature representation capacity. 3M-CDNet gains a good generalization capacity by incorporating effective “tricks” to alleviate overfitting, in which online data augmentation (Online DA) is applied to increase the diversity of the training samples, and Dropout regularization is applied in the classifier. Extensive ablation studies have proved the effectiveness of the core components. Experiment results suggest that 3M-CDNet outperforms state-of-the-art methods on several optical RS datasets and serves as a new universal benchmark. Specifically, 3M-CDNet achieves the best F1-score, i.e., LEVIR-CD (0.9161), Season-Varying (0.9473), and DSIFN (0.7031).

show abstract

Foreground-Aware Relation Network for Geospatial Object Segmentation in High Spatial Resolution Remote Sensing Imagery

Cited by 218 publications

References 47 publications

GourmetNet: Food Segmentation Using Multi-Scale Waterfall Features with Spatial and Channel Attention

GourmetNet: Food Segmentation Using Multi-Scale Waterfall Features with Spatial and Channel Attention

AGCDetNet:An Attention-Guided Network for Building Change Detection in High-Resolution Remote Sensing Images

3M-CDNet: A Universal Powerful Benchmark for Remote Sensing Image Change Detection

Contact Info

Product

Resources

About