Menu
In a time when the ecological effects of building abound, understanding is crucial; there exists a revolutionary system called CAPES, which stands for Classify Areas with Potential, then Exclude the Stable pixels, and promises to change how road construction alterations are supervised. This new method successfully overcame most challenges when mapping medium-sized changes in construction by using data from satellites over time and models for deep learning; these challenges include a broad range of spectral values during various stages of building development and small or isolated modification signals.
The application of deep convolutional layers has historically posed challenges in feature extraction. However, the CAPES method has effectively addressed this issue by integrating a U-net model with per pixel-based time series from the COLD algorithm, which continuously monitors land disturbance. This unique combination has shown significant improvements, with the U-net model, when boosted by a time-series coefficient and an RMSE value before and after adjustment, producing an impressive average F1-score of 70.8 %. This technique also minimizes the loss of spatial information related to minor, isolated changes in buildings.
The CAPES method, surpassing other pixel-based machine learning algorithms, not only monitors construction changes but also excels in monitoring construction changes. Its fine-tuning capability ensures unbiased monitoring across all land areas and dates. This advancement is set to significantly enhance road construction project monitoring and management, providing a more precise and effective tool for understanding and mitigating construction-related environmental impacts
Address: 270 George St N, Peterborough, ON K9J 3H1