Student research topics

At the chair of Digital Signal Processing and Circuit Technology student topics in the field of digital data transmission technologies, computerbased behavoir analysis and computer vision on embedded systems are available. The given topics can be modified to the personal interests of the student. Master and diploma thesises are assigned in coordination with the student. Here it is tried to find a solution that allows a continuation of the topic as a researcher at the chair after the completion of the studies.

Please send your request to request-dst@...

Individual topics could registrated according to prior agreement.

The following topics can be dealt with at the chair of Digital Signal Processing and Circuit Technology:

Topic	BT	RP	MT
Synthetic Data Acquisition		x
Optical Flow Ground Truth		x	x
Analyse und Verarbeitung omnidirektionaler Bilder	x		x
Entwicklung eines Klassifikators von Bildern	x
Combine Mesh Rigging and Skinning with 3D Skeleton Data			x
2D Image to 3D Model		x
Motion and edge fusion via dense extreme inception network			x
Unsupervised Domain Adaptation for Synthetic Images		x	x
Facial Recognition on Tablet Computers			x
3D Skeleton Reconstruction from Omnidirectional Image			x

The following topics are currently in progress at the chair of Digital Signal Processing and Circuit Technology:

Topic	Adviser	Type
Eye tracking: Improvement of prediction accuracy by the combination of classifiers	Jingrui Yu	MT
Investigation of gender differences for the interest in different regions of human faces in varying age groups	Jingrui Yu	MT
Unsupervised Domain Adaptation for Object Detection with RAPID Architecture	Ana Cecilia Perez Grassi, Tobias Scheck	MT
Performance comparison between anchor-based and anchorless architectures using synthetic images	Ana Cecilia Perez Grassi, Tobias Scheck	MT
Unsupervised Domain Adaptation for CenterNet using Fourier Domain Adaptation	Ana Cecilia Perez Grassi, Tobias Scheck	RP
Unsupervised Domain Adaptation for CenterNet using Adversarial Entropy Minimization	Ana Cecilia Perez Grassi, Tobias Scheck	RP
Instance Segmentation of Occluded Objects Using Deep Learning	Roman Seidel	MT
Evaluation of Domain Randomization with Synthetic Images using CenterNet	Ana Cecilia Perez Grassi, Tobias Scheck	RP
Dense Optical Flow Data Generation and Validation for Auto Encoder-Decoder Networks	Roman Seidel	MT
Saliency Maps Generation with Neural Networks	Ana Cecilia Perez Grassi, Tobias Scheck	RP
Self-Supervised Optical flow estimation with Fisheye camera	Roman Seidel	MT
CNN for Omnidirectional Depth Estimation	Julian Seuffert	MT
Efficient Deployment Of Recurrent Neural Networks On Embedded Systems	André Apitzsch	MT
Robotics process automation with artificial intelligence	Jingrui Yu	MT
Generation of Synthetic Data for Deep Learning in the Context of Activities of Daily Living	Tobias Scheck, Norbert Nestler	MT
Neural Network Runtime Optimization for ADAS and Autonomous Driving	Norbert Nestler	MT
Model transitions of object detection in different illumination scenarios	Roman Seidel	RP

The following topics were delt with at the chair of Digital Signal Processing and Circuit Technology in the last years:

Topic	Adviser	Type
Analysis and Comparison of Equalization Performance of OFDM and FBMC in Wireline Automotive Bus System	Sarah Cheraghinav	MT
Comparative Study between Camshift and CSRT Trackers on Omnidirectional Images	Ana Cecilia Perez Grassi	RP
Realtime Single Person 3D Human Pose Estimation using Deep Neural Network	Jyothsna Kondragunta	MT
Depth Estimation For Surround View Images using Object Detection and Ground - Wheel Contact Points	Gangolf Hirtz	MT
Efficient Neural Network training using small real world datasets	Jingrui Yu	MT
Quantization and Optimization of Machine Learning Models for Radar Based Object Classification	Ana Cecilia Perez Grassi	MT
CNN-based object detection at night	Roman Seidel	RP
Literature Research on CNNs for Stereo Vision	Julian Seuffert	RP
Training of Block Matching Parameters for 3D Indoor Scene Reconstruction	Julian Seuffert, Ana Cecilia Perez Grassi	RP
Class-Balanced Loss Based on Effective Number of Samples	Tobias Scheck, Ana Cecilia Perez Grassi	RP
Comparison of CNN based Semi Supervised Image Classification Methods	Ana Cecilia Perez Grassi, Tobias Scheck	RP
Human activity recognition from 3D skeletons	Jingrui Yu	RP
Influence of fisheye distortion for free space detection	Ana Cecilia Perez Grassi, Tobias Scheck	RP
Dynamic update of a 3D-point cloud using background subtraction on stereo images	Ana Cecilia Perez Grassi, Julian Seuffert	RP
Estimation of Gait Parameters based on 3D Human Gait	Jyothsna Kondragunta	RP
Domain Adaptation of Synthetic Generated Omnidirectional Images for Object Detection using Generative Adversarial Network	Tobias Scheck	MT
Domain Adaptation of Synthetic Generated Omnidirectional Images for Object Detection using Style Transfer Methods	Tobias Scheck	MT
Integration of two background models in order to identify new static objects in a scene	Ana Cecilia Perez Grassi	RP
Influence of Number of Classes for CNNs using Feature Space	Tobias Scheck, Ana Cecilia Perez Grassi	RP
Human activity recognition from 3D skeleton data using machine learning	Jingrui Yu	RP
Dense optical flow estimation and evaluation	Roman Seidel	RP
Design and Evaluation of a Miniaturized Flourescence Sensor Platform	Gangolf Hirtz	MT
Evaluation of Data Augmentation Methods to improve the Generalization of CNNs using Synthetic Data	Tobias Scheck, Ana Cecilia Perez Grassi	RP
3D Human Pose Estimation using Recurrent Neural Networks with Sequence-to-Sequence Architecture	Jyothsna Kondragunta	RP
A comparison of supervised and unsupervised optical flow estimation	Roman Seidel	RP
A survey to dense optical flow	Roman Seidel	RP
Selecting Influential Examples for Active Learning	Ana Cecilia Perez Grassi, Tobias Scheck	RP
Optical Flow Literature Review	Roman Seidel	RP
Implementierung eines Radarsensormodells in einer virtuellen Simulationsumgebung	Gangolf Hirtz	MT
Feature Space Dimension Reduction using PCA	Ana Cecilia Perez Grassi, Tobias Scheck	RP
Human Mesh Creation and Texturing with 3D sensors	Lars Lehmann	RP
Radio Resource Allocation Algorithms in Vehicle to Vehicle Communication for 5G Wireless Networks	Gangolf Hirtz	MT
Human Pose Estimation using 3D Skeleton Joint Position Data	Jingrui Yu	RP
Interactive Visual-Audio System for Fall detection in Assisted Living	Ana Cecilia Perez Grassi	MT
Influence of locally connected layers in tensorflow SSD on the quality of object detectors for omnidirectional top-view images	André Apitzsch	MT
Influence of locally connected layers in tensorflow faster R-CNN on the quality of object detectors for omnidirectional top-view images	André Apitzsch	MT
Online Object Detection using Lidar	Gangolf Hirtz	MT
Dataset of Synchronous Videos from Consecutive Vehicles using CARLA	Ana Cecilia Perez Grassi	RP
Sleep quality estimation based on video motion	Tobias Scheck	RP
Automatically learning the noise parameters ot the Kalman Filter	Gangolf Hirtz	MT
Particle Filter for omnidirectional images	Ana Cecilia Perez Grassi	RP
Fall detection on omnidirectional images	Ana Cecilia Perez Grassi	RP
Development of a Smart Home Sensor for Water Usage	Norbert Nestler	RP
Point Cloud Filtering for Plane Detection	Julian Seuffert	RP
Vergleich verschiedener Algorithmen zur Lokalisierung anhand der Empfangssignalstärke	Frank Klöpfel	BT
Setup and Evaluation of a Trinocular Omnidirectional Stereo Vision System for Indoor Surveillance	Julian Seuffert	MT
Daily Object and Human Pose detetion in Omnidirectional Camera Images Using Neural Networks	Jingrui Yu	MT
Parameter tuning of a convolutional neural network for optical flow determination	Roman Seidel	MT
Human Activity Recognition for both Predefined and Unseen Classes in the Context of Robots	Norbert Nestler	MT
Human body estimation in omnidirectional camera images using SSD and OpenPose	Jingrui Yu	MT
Synthetic Data Acquisition	André Apitzsch	RP
Analyse, Entwurf und Evaluierung von Verfahren zur Verfolgung mehrerer Personen in Bildsequenzen	Roman Seidel	BT
Analysis of automotive LiDAR sensor data, converting the point cloud data into 2D plots for visualization and performing classical object recognition	André Apitzsch	RP
From Points to Images: Transformation of single planes to binary images	Roman Seidel	RP
Human Pose Estimation by Maschine learning using TensorFlow	Jyothsna Kondragunta	RP
Deep learning based Human Pose Estimation using Theano	Jyothsna Kondragunta	RP
Implementation of an Enhanced Testbench based on the Open Source VHDL Verification Methodology	Gangolf Hirtz	RP
Object Detection from 3D Sensor Raw Data in Autonomous Driving	André Apitzsch	MT
Adaptive Anti-aliasing for fisheye cameras in surround view systems	Roman Seidel	MT
Design of a 200GHz distributed voltage-controlled Oszillatorv (DVCO)	Gangolf Hirtz	MT
3D-Plane-Estimation in Pointcloud and Depth-Data	Roman Seidel	RP
Stereo Correspondance on FPGA	André Apitzsch	RP
Digital Image Sharing by Diverse Image Media	André Apitzsch	RP
Object Detection using CUDA	André Apitzsch	MT
Plane Detection using Single Stereo Image	André Apitzsch	RP
Implementation of a Plane Detection for Stereo Vision	André Apitzsch	RP
Software Toolbox for a modular Vision System	André Apitzsch	RP
Implementierung einer OFDM basierten Ethernet Übertragung auf einem DSP über UTP	Gangolf Hirtz	DT
DC-DC-Wandler zur Bereitstellung einer Hilfsversorgung	Gangolf Hirtz	BT
Untersuchung von merkmalsbasierenden dynamischen Hintergrundmodellen	Gangolf Hirtz	BT
Automatisierte Bewertung und Darstellung von E/E-Architekturen im Kaftfahrzeug	Gangolf Hirtz	DT

BT - Bachelor Thesis, RP - Research Project, MT - Master Thesis, SP - Seminar Paper, DT - Diploma Thesis