Projects

ProApp/Aymatic Video (2yrs)

Android, C#, Xamarin, FFmpeg, Firebase

  • Guided video creation using Android app.
  • Full video recording, editing, sampling, rendering, video project management software which guides the user through the video creation process informed by expert knowledge (in video marketing).
  • Worked exceptionally well for event videos.

  • Recording Module:

    Custom Video Recorder guiding the user, with a transparent overlay to show the subject to be filmed and visual queues for timing the length of the shot (MediaRecorder, Camera, SurfaceTexture) Focus with custom focus rectangle and camera settings management by the app.

  • Player Module:

    Custom Video Player with accurate seeking (using ExoPlayer2 api) Live virtual cutting:, only the selected section of the timeline was played

  • Firebase Module:

    Connection to google firebase for authentication, user management, dowloading of selected content used by the application. Creation of objects based on online data using Factory Pattern.

  • Rendering Module:

    FFmpeg (native C++ library) custom built with ubuntu for use with Android (as a custom build was necessary to use a LGPL licence instead of the default GPL licence). FFmpeg was used to cut the video, add animations, remix music + audio and put everything together into a single video.

  • Preview Player:

    Playing a preview of the full video without rendering

  • Common Module:

    A PCL (Portable Class Library) Library for sharing models and behaviours between platforms.

  • Whole Project was refactored several times, where heavy emphasis on abstraction, modularization and reusability was set.

Hand Pose Recognition with Deep Neural Networks (1yr)

(Bachelor Thesis) Python, Tensorflow

Implementation of a Convolutional Neural Network with tensorflow and python. Goal was to classify the degree of hand contraction from hand images, by training a neural network for this task.

  • Created a framework for rapid testing of different datasets and models.
  • Created a dataset from videos and alternatively from generated data of a 3d model.
  • Trained Model and optimized result
  • Implemented real time prediction of images from a feed (webcam, video)

Aymatic Web (1yr)

Angular, NodeJs, Typescript, Javascript, Html, Css, FFmpeg

  • Automatic video creation using product data from websites.
  • Could take your shop and automatically create videos with music and visualization from your products and inserts those videos into your shop.
  • All of that fully automatically.
  • Manual selection of content was also possible (was used for videos of cars from a car reseller)

Power Mind Map

C#, UWP

  • Vector Graphics Mind Mapping App for Windows 8 and 10, implemented using C#, XAML and Universal Windows Platform UWP

Robotracking

C++, OpenCV

Implementation of a Convolutional Neural Network with tensorflow and python. Goal was to classify the degree of hand contraction from hand images, by training a neural network for this task.

  • Automatisation of a robot crane, utilizing 2 webcams.
  • The cranes purpose was to pick up a colored ball, the webcams and the sofware had to recognize the crane and the ball under the crane and send appropriate instructions to the robot to accomplish that task.

Custom Websites

HTML,CSS,JS,PHP