Выбрать из множества учителей по предмету английский...
Публикации
cover letter for Ph.D. position application (part 1) Dear Prof. Roth, My name is CZ and currently I am a graduate student at TUDa, majoring in computational enginnering (robotics). My research interests are focused on computer vision, including not only traditional geometrical and statistical methods, but also learning based methods. While doing my master thesis, I have implemented a 3D reconstruction method combining Multi-View Stereo and deep generative models. Instead of extracting features, I use VAE-GAN at the feature extraction phase so that less high level features will get lost. The proposed method has improved the baseline model greatly and has shown pretty good performances on indoor scene reconstruction. On top of that, I have also jointly implemented a stair detection algorithm for robot trajectory planning. This project helps me get familiar with non-linear optimization and different 3D representation forms like point cloud and TSDF map. Moreover, I designed and implemented a new approach for camera calibration as a team member, which is more efficient and advantageous than traditional methods. Besides researcher projects, I also have good machine learning skills. I completed several machine leaning, deep learning and computer vision courses at TU Da and got the best scores. Moreover, I also completed 7 different machine learning courses on Coursera, a online learning platform to improve my theoretical and practical skills. I am enthusiastic and motivated to do research at university. My bachelor- and masterthesis are both research-oriented. After finishing my bachelor thesis I have published 3 journal papers about vehicle engineering. My tutor and I are currently formulating a paper based on the result of my masterthesis. These experiences help me get into the research world and I personally pretty enjoy the whole research process. I think it is very exciting to go deep into a research area and contribute to the whole research community. During my current research internship in Munich, I am implementing a project about data shift in visual perception model in autonomous driving to make robuster models. I think the proposed Ph.D. positions are highly related to this. For example in Bayesian deep learning the parameters are not specific but obeying statistical distributions. By sampling the parameters Bayesian deep neural networks are more like generative approaches instead of discriminative approaches, which can be more potential to successfully deal with edge cases in many problems.
2 августа 2020 г.
1
0