Efficient Aerial Images Algorithms over Multi-objects labeling and Semantic Segmentation

Aerial imagery is gaining significance across various domains, comprising agriculture, disaster management, and surveillance. Durin this study, a unique method to tackle the persistent problems in accurate and efficient multi-object recognition in aerial imagery has been introduced. Our suggested approach combines the abstraction of spectral and spatial texture details with the incorporation of a color space segmentation algorithm. We convert the original data into a single feature space by utilizing multiple kernel learning (MKL), which improves feature representation accuracy and, in turn, improves classification accuracy. Finally, various objects in aerial images may be recognized and categorized according to the application of markov random field. By addressing variances in illumination, orientation, and size, this novel approach seeks to provide a complete solution for improving multi-object recognition in aerial data. Our findings demonstrate a significant improvement in object detection accuracy 98.13%.