| Summary: | Background: Geographical information about human settlements is necessary to facilitate humanitarian aid, recognize the location of real-world objects, support local development and improve disaster resilience in settlements. Although deep learning is employed in object location in settlements, these models have computational complexity and less detection accuracy. Hence they require further development with regards to precision in object location from geographic data. Objective: To efficiently identify the object location in Chinese settlements from geographic information, we proposed Artificial Neural Network (ANN) optimized by Red Fox (RF) optimization (ANN-RF) model in this research. Methods: The multi-resolution Geographic information data was collected from Tongzhou District, Beijing, China. ANN applies the measures of length, area, distance and direction to recognize object location from this data. An RF algorithm is used to optimize the weight vector of the ANN architecture to improve the efficiency of the ANN architecture. The parameters such as precision, recall, F1-score, number of correct detections, overall actual matches and the total number of detections were computed for the proposed model. Results: The precision rate and F1-score for the proposed ANN based RF optimization are about 98.9% and 97.5% which are higher than that of conventional models. Conclusion: The proposed ANN-RF model shows promising results in detecting object location in settlements from geographic information.
|
|---|