Jonas Geiselhart

Jonas Geiselhart

Sitemap

A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.

Pages

Posts

Detecting Objects from Aerial Images (2)

8 minute read

Published:

This article is about a project I did at the ISP at the University of Stuttgart. The second part is examining the transferability of our framework to other objects. We choose solar panel detection as such an other class, as there is a concrete demand to evaluate the development stages of renewable energies. Secondly solar panels exhibit a homogenous structure and are clearly visible and seperable, so an easy candidate for inference. In this case the goal is not to achieve the best inference results, but to test the transferability of our network.

Detecting Objects from Aerial Images (1)

15 minute read

Published:

This article is about a project I did at the ISP at the University of Stuttgart. The goal was to program a model and algorithm that detects & segments trees on a statewide scale in Southern Germany. While our focus is on the village areas, cities and forests should also achive feasible results. This article is split in two parts, the first one focuses on the development of the detection model and framework. In the second part I examine how this framework can be adapted to other targets, such as solar panel detection.

MOFormer - A Review

22 minute read

Published:

I did not do any research on this topic on my own, but only literature review, there is no content directly from me but only summary. If not directly cited the information is either considered common knowledge or from my references at the end, our main source that is reviewed is [1].

portfolio

Private Function Evaluation with Oblivious RAM

Published:

My bachelor thesis examines how much Private Function Evalutations, so evaluations computed encrypted by several parties, that mostly do not know the function and only a part of the data, can be simplified and accelerated by using Oblivious Random Access Machines, an encrypted storage.

Download here

Investigating Challenges in Generalizing Neural Radiance Fields with Learned Scene Priors

Published:

This thesis, conducted under the supervision of Shohei Mori at the Visualization Research Center (VISUS), investigates how neural radiance fields (NeRFs) can be extended from single-scene models toward generalizable representations. Such models promise major advantages in precomputability and applicability across autonomous systems, navigation, and immersive AR/VR environments.

Download here

publications

talks