Consultant Data Science/MachineLearning and Computer Vision

Gate state detector:
The goal of this project was to develop a deep learning model for infering the gate’s state (open,
closed, opening/closing) of a floodgate. My contribution was to retrain the model by collecting new
data for improvement in different weather conditions.
Tools: Tensorflow, ImageAI, OpenCV, python, Git
Parking control:
The goal of this project was to determine vehicle parking time to notify in case a car has parked
longer than the authorized time. This was done by vehicle detection, followed by licence plate
detection and recognition. My contribution on this project was the development of the licence plate
detection and recognition models.
Tools: Tensorflow, ImageAI, OpenCV, python, Git

Handwriting Optical Character Recognition
In this project, I develop an OCR model for extracting handwriting information from a form
Tools: Tensorflow, OpenCV, python
Traffic counting:
The goal of this project was to count vehicles passing on road using a camera and derive statistics
for each class of vehicle. For this purpose, I combined vehicle detection and tracking
Tools: Tensorflow, OpenCV, python, Git, Docker
People counting:
The goal of this project was to detect and count people passing on a street in both direction (left,
right) using a camera and derive statistics. For this purpose, I combined detection and tracking.
Data were then stored into a Postgres database and used to create visualizations on our analytic
platform using Superset
Tools: Tensorflow, ImageAI, OpenCV, python, Git, PostgresSQL, Superset

Parking space monitoring:
The goal of this project was to monitor a parking lot using cameras in order to determine if a parking
place is free or occupied at any time. My contribution to the project was to segment the vehicles
on the images using instance segmentation, compute the center of mass of the segmented vehicles,
then check if this is inside a parking place, for vehicle-to-parking place assignment.
Tools: Tensorflow, Mask-RCNN, OpenCV, python, Git, docker, Superset

Vorhersage des Energieverbrauchs:
Meine Aufgabe lag daran, die Time Series Daten des Energieverbrauchs der Kunden über die Jahre in Snowflake zu speichern, eines KI-Models zur Vorhersage des Energieverbrauchs von Kunden in Snowflake zu entwickeln, und ein Steamlit-basiertes Frontend zur Vorhersage ausgewählte Zeitfenster zu implementieren.
Tools: Python, Darts, Jupyter Notebook, git, Snowflake, Plotly

Online Road model for full autonomous driving
The goal of this project was to construct an online road model of the environment of the car by fusing
data from different sensor measurements. Dempster-Schafer-Theory is used to fuse lane markings,
semantic segmentation of drivable and non drivable area, occupancy grid, trajectory grid of other
participants. My contribution on this project was to improved the semantic segmentation, apply
Computer vision on the fused grid to extract geometrical lane information and develop Software
modules
Tools: Tensorflow, ROS, OpenCV, GitHub, Jira, Scrum, python, scikit learn, C++, CUDA

Tooling
My work on this project was to implement new KPIs and run them on measurements data for the
evaluation of autonomous driving scenarios, such as crossing and intersection, a stop traffic sign,
acceleration, braking, etc...
Tools: Jira, Scrum, python, pandas, matplotlib

I used Gaussian Mixture Models GMM, to predict movement of human by a robot arm in
industrial task, such as skewing for Human-Robot collaboration.
Tools: Python, C++, ROS