Florian Fervers

prof_pic.jpg

Fraunhofer IOSB

Karlsruhe, Germany

Welcome! I am a PhD student working at Fraunhofer Institute of Optronics, System Technologies and Image Exploitation (IOSB) in cooperation with the Computer Vision for Human-Computer Interaction Lab (CV:HCI) at Karlsruhe Institute of Technology (KIT) under supervision of Prof. Rainer Stiefelhagen. I completed my Masters in Computer Science at KIT in 2020 with an emphasis on machine learning and robotics. My current research focuses on GNSS-free self-localization for vehicles by exploiting globally available aerial imagery that is matched against the vehicle’s sensor readings.

News

Dec 14, 2023 Published the Python libraries einx and weightbridge to improve the implementation of deep learning models.
Jun 28, 2023 I have been invited to hold a talk on Cross-view Geolocalization at the 18th Dortmunder Autotag.
Apr 16, 2023 I have been invited to hold a talk on Cross-view Geolocalization at the CVPR 2023 tutorial A Comprehensive Tour and Recent Advancements toward Real-world Visual Geo-Localization.
Feb 27, 2023 Our paper Uncertainty-aware Vision-based Metric Cross-view Geolocalization has been accepted at CVPR 2023.
Jun 30, 2022 Our paper Continuous Self-Localization on Aerial Images Using Visual and Lidar Sensors has been accepted at IROS 2022.

Selected Publications

  1. Florian Fervers, Sebastian Bullinger, Christoph Bodensteiner, Michael Arens, and Rainer Stiefelhagen
    CVPR 2023
  2. Florian Fervers, Sebastian Bullinger, Christoph Bodensteiner, Michael Arens, and Rainer Stiefelhagen
    IROS 2022
  3. Improving Semantic Image Segmentation via Label Fusion in Semantically Textured Meshes
    Florian Fervers, Timo Breuer, Gregor Stachowiak, Sebastian Bullinger, Christoph Bodensteiner, and Michael Arens
    VISAPP 2022

Libraries

einx
Tensor Operations in Einstein-Inspired Notation for Python.
weightbridge 🌉
Map (deep learning) model weights between different model implementations in Python.

Semantic Meshes
Framework for annotating 3D meshes using the predictions of a 2D semantic segmentation model.
tinylogdir
Lightweight library for creating logging directories in python scripts.