Improved Visual Localization via Graph Filtering.

Vision-based localization is the problem of inferring the pose of the camera given a single image. One commonly used approach relies on image retrieval where the query input is compared against a database of localized support examples and its pose is inferred with the help of the retrieved items. This assumes that images taken from the same places consist of the same landmarks and thus would have similar feature representations. These representations can learn to be robust to different variations in capture conditions like time of the day or weather. In this work, we introduce a framework which aims at enhancing the performance of such retrieval-based localization methods. It consists in taking into account additional information available, such as GPS coordinates or temporal proximity in the acquisition of the images. More precisely, our method consists in constructing a graph based on this additional information that is later used to improve reliability of the retrieval process by filtering the feature representations of support and/or query images. We show that the proposed method is able to significantly improve the localization accuracy on two large scale datasets, as well as the mean average precision in classical image retrieval scenarios.

Thin debris layers do not enhance melting of the Karakoram glaciers.

The assessment of meltwater sourcing from the clean and debris-covered glaciers is scarce in High Mountain Asia (HMA). The melting rate varies with the debris cover thickness and glacier orientation. The present study quantifies glacier melting rate attributed to varying thickness of debris cover in the Karakoram. We observed daily melting rates by installing ablation stakes over debris-free and debris-covered ice during a field expedition. The stakes were installed on glacier surface with debris cover thickness ranges between 0.5 and 40 cm at selected experimental sites during the ablation period (September and October 2018) and (July to August 2019). We selected three glaciers including Ghulkin, Hinarchi, and Hoper facing east, south, and north, respectively to assess the role of glacier orientation on melting rates. We observed that the debris-free ice melts faster than the debris-covered ice. Intriguingly, a thin debris layer of 0.5 cm does not enhance melting compared to the clean ice which is inconsistent with the earlier studies. The melting rate decreases as the thickness of debris cover increases at all the three selected glaciers. Furthermore, south-facing glacier featured the highest melting (on average ~ 25% more). However, the north and east-facing glaciers revealed almost same melting rates. We observed that the average degree-day factors (DDF) slightly varies within a range of 0.58-0.73 and 0.55-0.68 cm °C-1 day-1 for debris-free and 0.5 cm debris-covered ice, respectively, however, DDF largely reduces to 0.13-0.25 cm °C-1 day-1 for 40 cm debris-covered ice. We suggest continuous physical glacier ablation observations for various debris cover throughout the ablation zone to better understand the role of debris on melting.