Supervisors: Marcin Grzegorzek, Lukas Köping, Kimiaki Shirahama
Numerous application requires position tracking system that works indoor environment.Object location can be identified by combination of different sensors data. The objective with sensor fusion is to combine measurements from different sensors to improve the quality of information such that the combined information becomes more valuable than the information from each individual sensor. JINS MEME a smart eye-wear, is a subtle and intriguing smart eye wear, with three basic sensors accelerometer, gyroscope(Inertial measurement unit sensor) and Electrocardiography (EOG/E.O.G.) is used for motion tracking of pedestrian by fusing sensor data with Extended Kalman filter.To strong hold the system, data from an android smart phone additionally consisting magnetometer (MARG) sensor has been integrated for indoor localization by Madwick filter.
Semester:
Summer Term 2016
Student:
Philip Gouverneur
Supervisor:
Marcin Grzegorzek, Lukas Köping, Kimiaki Shirahama
Emotion recognition is seen to be an important part in computer science. Machines will require at least some skills in emotion recognition to appear intelligent or to communicate useful feedback. But there are also other applications for emotion recognition than artificial intelligence. In the project 'Cognitive Village' recognition of emotions could be used to help old and sick people to live independent in their own homes as long as possible. Changes in the typical cycle of emotions could detect a disease or simply the need for help.
In this theses a wristband (Empatica E4 wristband) is used to collect data. The band contains several sensors, as a PPG Sensor, an Accelerometer and many more. Emotions got different impacts on the human body and therefore changes in the measured data. A codebook approach is used to classify this data, to recognize emotions in the last step.
Dear students, I would like to bring your kind attention to the following points:
The examination materials are from the posted slides, exercises and introduced references.
The share of the practical part will be around 30% and the rest will cover the discussed theory, methods and principles in the lectures.
The project presentations will take place on Friday 02.02.2018. Each group is given ca. 12 mins time slot for presentation.
The due of the first exercise set is 15.11.2017.
The due of the second exercise set is 29.11.2017.
Due to a bussines trip the next exercise class will take place on 20.12.2017. (Update:The project topics have been updated and they are discussed on 20.12.2017)
The third exercise will be uploaded shortly and it is due on 10.01.2018 (The new deadline is 17.01.2018).
Remark: After selecting your time slot please send me an Email to assign the exact examination time. Also please be advised that maximum number of attendees per 2-hour slot is 3. Thanks for your collaboration.
Recommended Book: Applied Medical Image Processing, A Basic Course, By: Wolfgang Birkfellner, CRC Press, 2011,
If you have already implemented the retrieval system, please try to extract additional edge features (HOG: Histogram of Oriented Gradients) by referring to [1] and [2]
Local Features (SIFT: Scale-Invariant Feature Transform)
18.01
Compared to the original slides I used at the lecture, these uploaded slides miss many images that may be related to copyright issues. If you want to have the original slides, please e-mail me.
Local Features: SIFT Feature Extraction by OpenCV
25.01
Local Features: Feature Matching
01.02
Additional Information:
Please take your own laptops to the exercise.
Programming skill for C is necessary for implementing the codes (C++ is desirable, but is not necessary).
If you want to use other programming languages like java and python, I will support you as much as possible.
In the oral examination, you will be asked several questions about the exercise course.
Semester:
Summer Term 2016
Student:
Philip Gouverneur
Supervisor:
Marcin Grzegorzek, Lukas Köping, Kimiaki Shirahama
Emotion recognition is seen to be an important part in computer science. Machines will require at least some skills in emotion recognition to appear intelligent or to communicate useful feedback. But there are also other applications for emotion recognition than artificial intelligence. In the project 'Cognitive Village' recognition of emotions could be used to help old and sick people to live independent in their own homes as long as possible. Changes in the typical cycle of emotions could detect a disease or simply the need for help.
In this theses a wristband (Empatica E4 wristband) is used to collect data. The band contains several sensors, as a PPG Sensor, an Accelerometer and many more. Emotions got different impacts on the human body and therefore changes in the measured data. A codebook approach is used to classify this data, to recognize emotions in the last step.
