Datasets:
Formats:
imagefolder
Size:
< 1K
ArXiv:
Tags:
visual-place-recognition
multimodal-learning
graph-neural-networks
urban-computing
pedestrian-navigation
day-night-recognition
License:
Update README.md
Browse files
README.md
CHANGED
|
@@ -43,18 +43,59 @@ tags:
|
|
| 43 |
|
| 44 |
### Dataset Statistics
|
| 45 |
|
| 46 |
-
| Modality | Count | Source | Details |
|
| 47 |
-
|----------|-------|--------|---------|
|
| 48 |
-
| **Images** | 110,529 | Field + Social Media | Resolution: 256×192 (preprocessed) |
|
| 49 |
-
| **Videos** | 2,527 | Field Collection | 20-60s, 256×144, 30fps |
|
| 50 |
-
| **
|
| 51 |
-
| **
|
|
|
|
|
|
|
| 52 |
|
| 53 |
**Spatial Composition:**
|
| 54 |
- 81 Nodes (street intersections)
|
| 55 |
- 125 Edges (61 horizontal + 64 vertical streets)
|
| 56 |
- 2 Squares (large open spaces)
|
| 57 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 58 |
## Dataset Structure
|
| 59 |
|
| 60 |
The dataset comprises four main components:
|
|
|
|
| 43 |
|
| 44 |
### Dataset Statistics
|
| 45 |
|
| 46 |
+
| Modality | Count/Coverage | Source | Details |
|
| 47 |
+
|----------|----------------|--------|---------|
|
| 48 |
+
| **Images** | 110,529 images | Field + Social Media | Resolution: 256×192 (preprocessed) |
|
| 49 |
+
| **Videos** | 2,527 clips | Field Collection | 20-60s, 256×144, 30fps |
|
| 50 |
+
| **Texts** | 208 locations | OCR + Manual | GPS, store names, signage, spatial metrics |
|
| 51 |
+
| **Graph Structure** | 208 nodes/edges | Urban Network | Connectivity, distances, space syntax |
|
| 52 |
+
| **Locations** | 208 unique places | Chengdu Taikoo Li | ~70,800 m² pedestrian district |
|
| 53 |
+
| **Temporal Coverage** | 7 years | 2019-2025 | Fine-grained + long-term span |
|
| 54 |
|
| 55 |
**Spatial Composition:**
|
| 56 |
- 81 Nodes (street intersections)
|
| 57 |
- 125 Edges (61 horizontal + 64 vertical streets)
|
| 58 |
- 2 Squares (large open spaces)
|
| 59 |
|
| 60 |
+
---
|
| 61 |
+
|
| 62 |
+
## Visual Overview
|
| 63 |
+
|
| 64 |
+
### Dataset Framework
|
| 65 |
+
|
| 66 |
+
Our dataset addresses four critical limitations in existing VPR datasets through systematic multimodal data collection:
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
|
| 70 |
+
*Figure 1: MMS-VPR framework addressing four VPR limitations: (1) pedestrian-only perspective, (2) day-night coverage, (3) multimodal integration (images + videos + text), and (4) 7-year temporal span enhanced with social media data.*
|
| 71 |
+
|
| 72 |
+
### Key Components Visualization
|
| 73 |
+
|
| 74 |
+
<table>
|
| 75 |
+
<tr>
|
| 76 |
+
<td width="50%">
|
| 77 |
+
|
| 78 |
+
**Data Collection Pipeline**
|
| 79 |
+
|
| 80 |
+
|
| 81 |
+
|
| 82 |
+
*Systematic methodology: site collection → data processing → textual annotation → social media integration*
|
| 83 |
+
|
| 84 |
+
</td>
|
| 85 |
+
<td width="50%">
|
| 86 |
+
|
| 87 |
+
**Benchmark Platform**
|
| 88 |
+
|
| 89 |
+
|
| 90 |
+
|
| 91 |
+
*MMS-VPRlib: Unified platform integrating datasets, models, and evaluation pipelines*
|
| 92 |
+
|
| 93 |
+
</td>
|
| 94 |
+
</tr>
|
| 95 |
+
</table>
|
| 96 |
+
|
| 97 |
+
---
|
| 98 |
+
|
| 99 |
## Dataset Structure
|
| 100 |
|
| 101 |
The dataset comprises four main components:
|