Authors: Deyi Ji, Siqi Gao, Mingyuan Tao, Hongtao Lu, Feng Zhao
Published on: December 29, 2023
Impact Score: 7.4
Arxiv code: Arxiv:2312.17428
Summary
- What is new: Introduction of ChangeNet, a large-scale dataset for multi-temporal change detection with 31,000 image pairs from 100 cities and coverage of a wide range of scenes and six pixel-level annotated categories.
- Why this is important: Existing change detection datasets are small, short in temporal scope, and low in practicability due to the high costs of multi-temporal images acquisition and labeling.
- What the research proposes: The creation of the ChangeNet dataset to improve the scope, scale, and practicality of change detection datasets, making it suitable for both binary and semantic change detection tasks.
- Results: Benchmarking on six binary change detection methods and two semantic change detection methods showcased the challenges and significance of the ChangeNet dataset.
Technical Details
Technological frameworks used: nan
Models used: Benchmarked using six BCD methods and two SCD methods
Data used: ChangeNet dataset with 31,000 multi-temporal images pairs from 100 cities, covering 6 pixel-level annotated categories
Potential Impact
Geospatial analysis, urban planning, environmental monitoring, and companies in satellite imaging and real-time surveillance markets.
Want to implement this idea in a business?
We have generated a startup concept here: TempoTrace.
Leave a Reply