REFERENZEN

COMPUTER VISION (CV)

Branche: Hi-Tech - Unternehmensgröße: <200

Projektinhalt:

Entwicklung effizienter Bildverarbeitungsalgorithmen einschließlich Implementierung, Testen und Dokumentation, Vollständige Portierung des internen Buildsystems von Make nach CMake, Weiterentwicklung und Pflege bestehender Algorithmen, insbesondere Optimierung und Parallelisierung, Cross-Plattform Entwicklung für Windows, macOS, Linux sowie Embedded-Geräte.

Technologien & Methodik:

C, Doxygen, Bildverarbeitung, HALCON, Qt, Python, Caffe, Make, CMake, Jenkins, Perl, OpenMP, OpenCL, SIMD, SSE, AVX

Branche: IT - Unternehmensgröße: <100

Projektinhalt:

Forschung und Entwicklung eines Verfahrens zur fälschungssicheren, eindeutigen Erkennung von Krakelee-Mustern auf Basis von Bildmerkmalen und Oberflächenstrukturen.

Technologien & Methodik:

C++, Python, Boost CMake Qt OpenGL VTK OpenCV Doxygen FlyCapture libdc1394 OpenMP

Branche: Hi-Tech - Unternehmensgröße: <200

Projektinhalt:

Development and optimization of computer vision algorithms for machine vision. Mainly in the areas deep learning, identification (Bar-/Datacode), machine learning, and core image processing functionalities (Filters, Thresholds, Segmentation, Computational Geometry, etc.). Rewrite of the 2D visualization. 3rd-Level customer support for a wide range of complex 3D and 2D computer vision and machine learning problems.

Design and improvements of the build, test and CI infrastructure. Refactoring, modularization and redesign of a large legacy C codebase. Cross-platform development for Windows, Linux, macOS, and embedded devices. Porting the machine vision library to Android.

Technologien & Methodik:

C, cuDNN, cuBLAS, Intel MKL, Skia, C++, CMake, Jenkins, Scrum, Kanban

Branche: Automotive - Unternehmensgröße: <4.500

Projektinhalt:

Design, development, and implementation of a real-time traffic sign recognition system for an Android/iOS app (ACoDriver).

Technologien & Methodik:

OpenCV, Java, C/C++ Neural Networks, Local binary patterns, Boosting

Branche: Automotive - Unternehmensgröße: <10.000

Projektinhalt:

Improving lane detection and lane-departure warning system for ADAS. Training and tuning the algorithms to work with hard road scenarios. Team achieved >96% availability (92% required).

Technologien & Methodik:

image processing, image features, pattern recognition, object detection and tracking, OpenCV, Python, C++, Tensorflow, Keras, CNN, SVM

Branche: Start-Up - Unternehmensgröße: <50

Projektinhalt:

Development of a mobile application for video life blogging with VR and AR elements. Face/body detection and tracking up to 4m with smart-phone camera. Image stitching to panorama from video frames.

Technologien & Methodik:

image processing, image features, pattern recognition, object detection and tracking, decision trees, OpenCV, Python, C++, Matlab, Tensorflow, Keras, DNN, CNN

Branche: Maschinenbau - Unternehmensgröße: <3.500

Projektinhalt:

Industrielle Bildverarbeitung, Prototypenentwicklung und Algorithmenentwicklung im Bereich Industrielle Bildverarbeitung – automatische Konturenerkennung, Implementierung einer semi-automatischen Korrekturfunktion.

Technologien & Methodik:

C++, OpenCV, Konturenerkennung, Graph Cut

Branche: Start-Up - Unternehmensgröße: <35

Projektinhalt:

Algorithmenentwicklung zur Gesichtserkennung: Bildvorverarbeitung, Gesichtsdetektion, Analyse der Textureigenschaften, Identifizierung, Verifizierung; Optimierung: – Parameteranpassung mit numerischen Methoden, Normalisierung ungleichmäßiger Beleuchtung an Gesichtsbildern.

Technologien & Methodik:

C++, OpenCV, MATLAB, Max-Flow Min-Cut, Graphentheorie, Extraktion von grob- und fein-skalierten Merkmalen, Segmentierung, Principal Component Analysis (PCA), Logistic Regression, Linear Regression, Gradient Descent

Branche: Automotive - Unternehmensgröße: <10.000

Projektinhalt:

ADAS – Algorithmenentwicklung und Codeoptimierung im Bereich Fahrerassistenzsysteme / Laserscanner. Prototypenentwicklung und Algorithmenentwicklung im Bereich Fahrerassistenzsysteme – Kamerakalibrierung, Fahrspurerkennung, Implementierung von real time PC-basierten Prototypen im Fahrzeug.

Technologien & Methodik:

C++, MATLAB, Python, Jupyter Notebooks, Keras, Tensorflow, Convolutional Networks,Bayesian Networks, Tracking, Klassifizierung

Branche: MedTech - Unternehmensgröße: <10.000

Projektinhalt:

3D Reconstruction from Multiple views in Matlab, C++.

Technologien & Methodik:

Matlab, C++, OpenCV, OpenMP

Branche: Marketing - Unternehmensgröße: <300

Projektinhalt:

Building a prescriptive model for advertisement text (with revenue uplift 7%) – conducted feasibility study for user recommendations based on recurrent neural networks.

Technologien & Methodik:

Python, TensorFlow, RNN, Machine & Deep Learning

Branche: Marketing - Unternehmensgröße: <300

Projektinhalt:

Text Mining: implementation of a new, scalable item based collaborative filtering algorithm for 150 million events per day.

Increase in click through rate by 10% compared to previous implementation – constructed real time streaming pipeline to transform newspaper text into vector format for about 20 texts per second being transformed, inserted and updated each day, implemented a fully automated model retraining pipeline.

Technologien & Methodik:

Python, Spark, TensorFlow, Machine & Deep Learning, Big Data

Branche: Banken - Unternehmensgröße: <1.200

Projektinhalt:

Design and implement a Data Science Stack for Text Mining.

Technologien & Methodik:

Regression, Classification and Apache Spark, Recurrent Neural Networks, Keras, Hadoop

Branche: Medien - Unternehmensgröße: <5.500

Projektinhalt:

Topic modeling: Use LDA-Algorithm to model the topics of German news
Doc2Vec: Use the topic model to transform each document to a topic vector
Similar documents detection: Use the topic model to transform the corpus to a topic matrix and search similar documents using the corpus-topic matrix
Keyword identification: Build a keyword extractor to identify keywords from German news

Technologien & Methodik:

Python, Gitlab, Jmeter, SonarQube, Docker, Rancher

Branche: E-Commerce - Unternehmensgröße: <1.000

Projektinhalt:

Analyze the text of product names, descriptions, etc. to improve search quality, develop novel POS and entity tagger and build a prediction model of the product categories by training a feedforward neural network.

Technologien & Methodik:

Python, NLTK, Sklearn, machine learning, Neural Networks

Branche: Versicherung - Unternehmensgröße: <1.200

Projektinhalt:

Web-mining & Text-analytics: crawl websites, apply text analytics techniques to extract information.
Natural language processing for German: POS-tagging and stemming based on statistical inference; topic and sentiment analysis of the news.

Technologien & Methodik:

R, R-Shiny, IBM Watson

Branche: Retail - Unternehmensgröße: <4.000

Projektinhalt:

Development of reliable, robust and scalable machine learning systems and platforms, created 360° customer insights data platform that feeds data into several downstream systems including a DMP for online advertising, designed a privacy-preserving data science platform for marketing departments of 8 group companies to comply with changes in privacy laws and managed its implementation, developed solutions to create a common latent feature representation from heterogeneous internal and external data sources using novel deep learning techniques. Conducted internal “business development” to promote group wide use of services and data pools.

Technologien & Methodik:

Python, R, Hadoop, Spark, Pandas, Scikit-Learn, Keras, TensorFlow, PyMC3, R-Studio, PostgreSQL, MySQL, Vertica, Greenplum, Jupyter, Anaconda, Git, Jenkins

Branche: Gaming - Unternehmensgröße: <1.000

Projektinhalt:

Full life-cycle data science project management and machine learning system development. Professionalized development processes in start-up environment and replaced failure-prone systems with reliable and maintainable solutions, redesigned a marketing attribution system and increased accuracy through new predictive models and by connecting external data sources, helped to save significant affiliate marketing expenses by providing fraud management team with novel fraud detection algorithms, developed a robust and accurate model for customer lifetime value prediction that became part of the main Business.

Intelligence ETL pipeline and supported automated and manual decision making processes in several departments.

Technologien & Methodik:

Python, R, Hive, Pandas, Scikit-Learn, Keras, TensorFlow, PostgreSQL, MySQL

Branche: Pharma - Unternehmensgröße: <250

Projektinhalt:

Quantitative quality assessment of the company’s next generation sequencing (NGS) laboratory for in vitro diagnostic (IVD) accreditation, utilizing machine learning for collaboration projects and product innovation (eg, cancer subtype classication), designing a new data model, including genomic variants and cancer type taxonomy, as basis for research and products, conceptualizing and performing statistical analysis of high-dimensional observational data („big data“).

Technologien & Methodik:

R, Python, SQL, Matlab, Java, machine learning, text mining, statistical optimization

Branche: Telekommunikation - Unternehmensgröße: <9.000

Projektinhalt:

Geo-Location: Home location detection and occupational segments identification, Movement patterns for different segments, Visualization of geolocation and relevant information.

Technologien & Methodik:

Python (Pandas, Keras, nltk), Spark, R

Branche: Energiehandel - Unternehmensgröße: <750

Projektinhalt:

Modellierung von Finanzmarktrisiken

Technologien & Methodik:

Databricks on Azure, Spark Scala, Monte-Carlo-Simulation

Branche: Haushaltsgeräte - Unternehmensgröße: <2.500

Projektinhalt:

Aufbau einer Big-Data Analytics Abteilung durch Use Cases: Churn Analysis, B2B-Sales Optimization, Predictive Maintenance, Data quality.

Technologien & Methodik:

AWS, Spark, Scala, Sagemaker, TensorFlow, Keras, Python, RNNs

Branche: Elektronik - Unternehmensgröße: <2.400

Projektinhalt:

Design and development of a deep learning based sky/cloud classification system.

Technologien & Methodik:

Keras/Tensorflow, Python, AWS Cloud

Branche: Automotive - Unternehmensgröße: <20.000

Projektinhalt:

Design and implement advanced CNN architectures and models based on a huge number of Images for Shop Floor Control. Evaluate Deep Learning Architectures, Performance Optimization of Deep Learning Models.

Technologien & Methodik:

Image segmentation, Image classification, Anomaly detection, Video Analysis, Tensorflow and Keras with GPU (NVIDIA), Hadoop, GPU, PyCharm

Branche: Retail - Unternehmensgröße: <3.000

Projektinhalt:

Design, implement and optimize complex Machine Learning Systems for Online Marketing / eCommerce.

Technologien & Methodik:

Hive, Spark, Python, Deep Learning, K-Nearest Neighbors (KNN), SVM, Kernel SVM, Logistic Regression

Branche: Medien - Unternehmensgröße: <5.500

Projektinhalt:

Deep-Learning (NLP und Bild-/Videoverarbeitung)

Softwareentwicklung und Beratung für verschiedene Dienste einer Bild-Mining Software, Extractive Text Summarization / Image Captioning, (Human) Action Recognition, Object Detection.

Technologien & Methodik:

Python, TensorFlow, Kerasa, sk-learn, OpenCV, CNN

Branche: Hi-Tech - Unternehmensgröße: <200

Projektinhalt:

Entwicklung effizienter Bildverarbeitungsalgorithmen einschließlich Implementierung, Testen und Dokumentation, Vollständige Portierung des internen Buildsystems von Make nach CMake, Weiterentwicklung und Pflege bestehender Algorithmen, insbesondere Optimierung und

Parallelisierung, Cross-Plattform Entwicklung für Windows, macOS, Linux sowie Embedded-Geräte.

Technologien & Methodik:

C, Doxygen, Bildverarbeitung, HALCON, Qt, Python, Caffe, Make, CMake, Jenkins, Perl, OpenMP, OpenCL, SIMD, SSE, AVX

Branche: IT - Unternehmensgröße: <100

Projektinhalt:

Forschung und Entwicklung eines Verfahrens zur fälschungssicheren, eindeutigen Erkennung von Krakelee-Mustern auf Basis von Bildmerkmalen und Oberflächenstrukturen.

Technologien & Methodik:

++, Python, Boost CMake Qt OpenGL VTK OpenCV Doxygen FlyCapture libdc1394 OpenMP

Branche: Hi-Tech - Unternehmensgröße: <200

Projektinhalt:

Development and optimization of computer vision algorithms for machine vision. Mainly in the areas deep learning, identification (Bar-/Datacode), machine learning, and core image processing functionalities (Filters, Thresholds, Segmentation, Computational Geometry, etc.). Rewrite of the 2D visualization. 3rd-Level customer support for a wide range of complex 3D and 2D computer vision and machine learning problems.

Design and improvements of the build, test and CI infrastructure. Refactoring, modularization and redesign of a large legacy C codebase. Cross-platform development for Windows, Linux, macOS, and embedded devices. Porting the machine vision library to Android.

Technologien & Methodik:

C, cuDNN, cuBLAS, Intel MKL, Skia, C++, CMake, Jenkins, Scrum, Kanban

Branche: Automotive - Unternehmensgröße: <4.500

Projektinhalt:

Design, development, and implementation of a real-time traffic sign recognition system for an Android/iOS app (ACoDriver).

Technologien & Methodik:

OpenCV, Java, C/C++ Neural Networks, Local binary patterns, Boosting

Branche: Automotive - Unternehmensgröße: <10.000

Projektinhalt:

Improving lane detection and lane-departure warning system for ADAS. Training and tuning the algorithms to work with hard road scenarios. Team achieved >96% availability (92% required).

Technologien & Methodik:

image processing, image features, pattern recognition, object detection and tracking, OpenCV, Python, C++, Tensorflow, Keras, CNN, SVM

Branche: S>tart-Up - Unternehmensgröße: <50

Projektinhalt:

Development of a mobile application for video life blogging with VR and AR elements. Face/body detection and tracking up to 4m with smart-phone camera. Image stitching to panorama from video frames.

Technologien & Methodik:

image processing, image features, pattern recognition, object detection and tracking, decision trees, OpenCV, Python, C++, Matlab, Tensorflow, Keras, DNN, CNN

Branche: Maschinenbau - Unternehmensgröße: <3.500

Projektinhalt:

Industrielle Bildverarbeitung, Prototypenentwicklung und Algorithmenentwicklung im Bereich Industrielle Bildverarbeitung – automatische Konturenerkennung, Implementierung einer semi-automatischen Korrekturfunktion.

Technologien & Methodik:

C++, OpenCV, Konturenerkennung, Graph Cut

Branche: Start-Up - Unternehmensgröße: <35

Projektinhalt:

Algorithmenentwicklung zur Gesichtserkennung: Bildvorverarbeitung, Gesichtsdetektion, Analyse der Textureigenschaften, Identifizierung, Verifizierung; Optimierung: – Parameteranpassung mit numerischen Methoden, Normalisierung ungleichmäßiger Beleuchtung an Gesichtsbildern

Technologien & Methodik:

C++, OpenCV, MATLAB, Max-Flow Min-Cut, Graphentheorie, Extraktion von grob- und fein-skalierten Merkmalen, Segmentierung, Principal Component Analysis (PCA), Logistic Regression, Linear Regression, Gradient Descent

Branche: Automotive - Unternehmensgröße: <10.000

Projektinhalt:

ADAS – Algorithmenentwicklung und Codeoptimierung im Bereich Fahrerassistenzsysteme / Laserscanner. Prototypenentwicklung und Algorithmenentwicklung im Bereich Fahrerassistenzsysteme – Kamerakalibrierung, Fahrspurerkennung, Implementierung von real time PC-basierten Prototypen im Fahrzeug.

Technologien & Methodik:

C++, MATLAB, Python, Jupyter Notebooks, Keras, Tensorflow, Convolutional Networks,Bayesian Networks, Tracking, Klassifizierung

Branche: MedTech - Unternehmensgröße: <10.000

Projektinhalt:

3D Reconstruction from Multiple views in Matlab, C++

Technologien & Methodik:

Matlab, C++, OpenCV, OpenMP

Branche: Hi-Tech - Unternehmensgröße: <750

Projektinhalt:

Multi-View ORB-SLAM.

Technologien & Methodik:

Python, C++, OpenCV, sk-learn, SLAM

Branche: Automotive - Unternehmensgröße: <7.500

Projektinhalt:

Sensor Fusion in Autonomous Driving – Integration sensor fusion in CarMaker simulation environment.

Technologien & Methodik:

C++, OpenCV, OpenML, Segmantation

Branche: Gaming - Unternehmensgröße: <1.000

Projektinhalt:

Full life-cycle data science project management and machine learning system development. Professionalized development processes in start-up environment and replaced failure-prone systems with reliable and maintainable solutions, redesigned a marketing attribution system and increased accuracy through new predictive models and by connecting external data sources, helped to save significant affiliate marketing expenses by providing fraud management team with novel fraud detection algorithms, developed a robust and accurate model for customer lifetime value prediction that became part of the main Business.

Intelligence ETL pipeline and supported automated and manual decision making processes in several departments.

Technologien & Methodik:

Python, R, Hive, Pandas, Scikit-Learn, Keras, TensorFlow, PostgreSQL, MySQL

Branche: Pharma - Unternehmensgröße: <2.500

Projektinhalt:

Inventing a method to prioritize genomic variants for clinical decision support, optimization of predictive multivariate biomarker from phase II clinical study data (developing a regularized proportional hazard model), providing statistical methodology for analysis of adverse events spontaneous reporting data (FDA FAERS).

Technologien & Methodik:

R, Python, SQL, Matlab, C++, machine learning, text mining, predictive analytics, statistical optimization

Branche: Pharma - Unternehmensgröße: <1.300

Projektinhalt:

Automated customer forecasting for workforce planning & customer satisfaction improvement.

Development of a machine learning model for Insurance Tenders to optimize inventory management, forecast sales of tenders, optimize the bycatch in pharmacies

Technologien & Methodik:

Python, Pandas, NumPy, TensorFlow, Machine Learning, statistical analytics

Branche: Retail - Unternehmensgröße: <4.000

Projektinhalt:

Annual sales planning to optimize personnel deployment plan, inventory management, advertising campaign planning, bonus calculation.

Technologien & Methodik:

Python, R, R-studio, TensorFlow, machine learning, advanced analytics

Branche: Versicherung - Unternehmensgröße: <3.500

Projektinhalt:

Aktien-Vorhersage mittels Long-Short-Term-Memory Neural Networks.

Technologien & Methodik:

Python, LSTMs, SVM, KNN Regressor, Random Forest Regressor

Branche: Versicherung - Unternehmensgröße: <10.000

Projektinhalt:

Exploit data and machine learning methods to develop the next-gen recommendations products. Predictive modeling for churn, next best product, risk scoring and financial systems.

Technologien & Methodik:

Python, R, NoSQL, Hadoop