Image Geo-Site Estimation Using Convolutional Auto-Encoder and Multi-Label Support Vector Machine

The estimation of an image geo-site solely based on its contents is a promising task. Compelling labelling relies heavily contextual information, which not as simple recognizing single object in image. An Auto-Encode-based support vector machine approach proposed this work to estimate the address issue misclassifying estimations. method for conducted using dataset consisting 125 classes various images captured within countries. uses convolutional Auto-Encode training and dimensionality reduction. After that, acquired preprocessed input further processed by multi-label machine. performance assessment has been accomplished accuracy, sensitivity, specificity, F1-score evaluation parameters. Eventually, presented article outperforms K-Nearest Neighbor Auto-Encode-Random Forest methods.