HCGMNET: A Hierarchical Change Guiding Map Network For Change Detection
Very-high-resolution (VHR) remote sensing (RS) image change detection (CD) has been a challenging task for its very rich spatial information and sample imbalance problem. In this paper, we have proposed a hierarchical change guiding map network (HCGMNet) for change detection. The model uses hierarchical convolution operations to extract multiscale features, continuously merges multi-scale features layer by layer to improve the expression of global and local information, and guides the model to gradually refine edge features and comprehensive performance by a change guide module (CGM), which is a self-attention with changing guide map. Extensive experiments on two CD datasets show that the proposed HCGMNet architecture achieves better CD performance than existing state-of-the-art (SOTA) CD methods.
PDF AbstractCode
Tasks
S2Looking
| Task | Dataset | Model | Metric Name | Metric Value | Global Rank | Benchmark |
|---|---|---|---|---|---|---|
| Change Detection | CDD Dataset (season-varying) | HCGMNet | F1-Score | 95.07 | # 12 | |
| Precision | 93.84 | # 3 | ||||
| F1 | 95.07 | # 3 | ||||
| Overall Accuracy | 98.82 | # 4 | ||||
| IoU | 90.60 | # 3 | ||||
| Recall | 96.34 | # 3 | ||||
| KC | 94.40 | # 2 | ||||
| Change Detection | DSIFN-CD | HCGMNet | F1 | 55.00 | # 7 | |
| IoU | 37.93 | # 6 | ||||
| Overall Accuracy | 76.26 | # 6 | ||||
| Precision | 40.57 | # 4 | ||||
| Recall | 85.35 | # 1 | ||||
| KC | 41.53 | # 4 | ||||
| Change Detection | GoogleGZ-CD | HCGMNet | F1 | 85.71 | # 2 | |
| Precision | 84.25 | # 3 | ||||
| Recall | 87.22 | # 2 | ||||
| Overal Accuracy | 92.85 | # 3 | ||||
| KC | 80.94 | # 3 | ||||
| IoU | 74.99 | # 3 | ||||
| Change Detection | LEVIR+ | HCGMNet | F1 | 82.37 | # 2 | |
| Prcision | 82.81 | # 1 | ||||
| Recall | 81.94 | # 2 | ||||
| OA | 98.57 | # 2 | ||||
| KC | 81.63 | # 2 | ||||
| IoU | 70.03 | # 2 | ||||
| Change Detection | LEVIR-CD | HCGMNet | F1 | 91.77 | # 11 | |
| IoU | 84.79 | # 8 | ||||
| Overall Accuracy | 99.18 | # 4 | ||||
| F1-score | 91.77 | # 3 | ||||
| Precision | 92.96 | # 4 | ||||
| Recall | 90.61 | # 5 | ||||
| Change Detection | S2Looking | HCGMNet | F1-Score | 63.87 | # 5 | |
| Precision | 72.51 | # 2 | ||||
| Recall | 57.06 | # 2 | ||||
| OA | 99.22 | # 1 | ||||
| KC | 63.48 | # 2 | ||||
| IoU | 46.91 | # 2 | ||||
| F1 | 63.87 | # 1 | ||||
| Change Detection | SYSU-CD | HCGMNet | F1 | 79.76 | # 3 | |
| Precision | 86.28 | # 2 | ||||
| Recall | 74.15 | # 4 | ||||
| OA | 91.12 | # 2 | ||||
| KC | 74.11 | # 2 | ||||
| IoU | 66.33 | # 2 | ||||
| Change Detection | WHU-CD | HCGMNet | F1 | 92.08 | # 4 | |
| Overall Accuracy | 99.45 | # 3 | ||||
| Precision | 93.93 | # 3 | ||||
| Recall | 90.31 | # 3 | ||||
| KC | 91.80 | # 3 | ||||
| IoU | 85.33 | # 3 |
