Results
Theses
2022
| Student | Title | Type | Location | Link | |
|---|---|---|---|---|---|
| Lukas Steindl | Optimizing deep neural networks for efficient dronebased ragweed detection | Master | TU Wien | https://doi.org/10.34726/hss.2022.79705 | |
| Nikolas Alge | Power Profiling of Machine Learning Accelerators using MLPerf | Bachelor | TU Wien | https://doi.org/10.5281/zenodo.7323206 |
2021
| Student | Title | Type | Location | Link | |
|---|---|---|---|---|---|
| Andreas Glinserer | Autopruning mit Intel Distiller und Evaluation auf einem Jetson Xavier AGX | Master | TU Wien | https://doi.org/10.34726/hss.2021.90301 | |
| Bernhard Haas | Compressing MobileNet With Shunt Connections for NVIDIA Hardware | Master | TU Wien | https://publik.tuwien.ac.at/files/publik_295948.pdf | |
| Michael Opitz | Efficient Ensembles for Deep Learning | Ph.D | TU Graz | ||
| Matvey Ivanov | Embedded Machine Learning Demonstrator | Bachelor | TU Wien | https://publik.tuwien.ac.at/files/publik_296007.pdf | |
| Dominik Dallinger | FPGA optimized dynamic post-training quantization of TinyYoloV3 | Bachelor | TU Wien | https://publik.tuwien.ac.at/files/publik_296008.pdf | |
| Amid Mozelli | A Study on Confidence: an Unsupervised Multi-Agent Machine Learning Experiment | Bachelor | TU Wien | ||
| Lukas Baischer | FPGA Based Embedded Neural Network Object Detector | Master | TU Wien | https://doi.org/10.34726/hss.2021.69314 |
2020
| Student | Title | Type | Location | Link | |
|---|---|---|---|---|---|
| Kaleab Alemayehu Kinfu | Lifelong Learning for Autonomous Vehicles: Monocular Depth Estimation | Master | TU Graz | ||
| Anam Zahra | Autonomous Vehicle Self-localization in Noisy Environments | Master | TU Graz | ||
| Marco Wuschnig | Auswertung verschiedener Methoden der Hyperparameteroptimierung in Machine Learning | Bachelor | TU Wien | https://publik.tuwien.ac.at/files/publik_295896.pdf | |
| Julian Roth | Auswertung von Cloudbasierten Machine Learning Frameworks für Supervised Machine Learning | Bachelor | TU Wien | https://publik.tuwien.ac.at/files/publik_295897.pdf | |
| Rudolf Wörndle | Continual Domain-Incremental Learning for Object Detection | Master | TU Graz |
Software
| Title | Description | Link |
|---|---|---|
| Blackthorn | Latency Estimation Toolkit for Neural Networks (Nvidia) | https://github.com/embedded-machine-learning/blackthorn |
| Embedded Machine Learning Scripts and Guides | We collect scripts and guides that help us in our everyday work to setup software and frameworks. This repository is also the source of an EML Toolbox that aims to easyily implement machine learning toolchains. | https://github.com/embedded-machine-learning/scripts-and-guides |
| ANNETTE | Accurate Neural Network Execution Time Estimation | https://github.com/embedded-machine-learning/annette |
| Re-Implementation of SqueezeNas | SqueezeNAS Repository with reimplementation for a custom semenatic segementation task | https://github.com/embedded-machine-learning/squeezenas_train |
| EML Mobile Phone Photo Detection Application | A mobile phone application for applying detection networks on captured fotos and camera streams | https://github.com/embedded-machine-learning/eml-mobile-photo-app |
| Shunt Connector | Neural network compression technique called 'Shunt connection' using Keras and TensorFlow 2.x as its backend. | https://github.com/embedded-machine-learning/ShuntConnector |
| MobileNetV3-Segmentation-Keras | Semantic segmentation version of the MobileNetV3 architecture (source), which is inspired by the DeeplabV3 architecture. The model is implemented using Keras and TensorFlow 2.x. | https://github.com/embedded-machine-learning/MobileNetV3-Segmentation-Keras |
Publications
See the Publication section.