Geological and scenic values of locations are the non-living curiosities that can be preserved and popularized a lot easier using the institutional background of geotourism, such as geoparks. UNESCO Global Geoparks Network is responsible for protecting and fostering natural, scenic and cultural values and especially geosites that are the exciting visible physical elements. Our goal was to quantify the geotourism potential around Csopak, a scenic village in the Balaton Uplands giving home for the headquarter of the Bakony-Balaton UNESCO Global Geopark. After designating 216 potential geosites using topographic and geological maps, we applied two assessment models: the Geosite Assessment Model (GAM) and the Modified Geosite Assessment Model (M-GAM). GAM has been applied with good results in Hungary on different areas, but M-GAM has not been used before. As M-GAM involves tourists into the process counting with their opinion, it may give a more realistic view of the geosites. The two methods produced different but comparable final values of geotourism potential counted from the Main Value and Additional Value scores. We discovered that the proportion of the difference of these values carries major information. The ratio of ΔAV/ΔMV used as linear functions and depicted on diagrams can derive which values are more important for the visitors. From this result we can draw conclusions about the future development trends: scientific or infrastructural values should be more effectively fostered. Using our results, geosites can be handled and developed as visitors expect it.
The main goal of the present research is to classify images of plants to species with deep learning. We used convolutional neural network architectures for feature learning and fully connected layers with logsoftmax output for classification. Pretrained models on ImageNet were used, and transfer learning was applied. In the current research image sets published in the scope of the PlantCLEF 2015 challenge were used. The proposed system surpasses the results of all top competitors of the challenge by 8% and 7% at observation and image levels, respectively. Our secondary goal was to satisfy the users’ needs in content-based image retrieval to give relevant hits during species search task. We optimized the length of the returned lists in order to maximize MAP (Mean Average Precision), which is critical to the performance of image retrieval. Thus, we achieved more than 50% improvement of MAP in the test set compared to the baseline.