Bernd Radig - Academia.edu (original) (raw)
Papers by Bernd Radig
Die Wissensrepräsentationssprache OPS5, 1987
Das ursprungliche in LISP implementierte OPS5 wird von einer LISP-Umgebung aus geladen. Es besteh... more Das ursprungliche in LISP implementierte OPS5 wird von einer LISP-Umgebung aus geladen. Es besteht weiterhin ein Zugriff auf das LISP-System. Dadurch wird einerseits der Aufruf externer (LISP-) Routinen etwas vereinfacht (die Unterstutzungsroutinen sind LISP-Funktionen), andererseits werden aber bei Laufzeitfehlern sowohl Fehlermeldungen des OPS5-Systems wie auch die des zugrundeliegenden LISP-Systems ausgegeben. Fur einen OPS5-Anwender sind daher Kenntnisse von LISP unumganglich.
Die Wissensrepräsentationssprache OPS5, 1987
Die Wissensrepräsentationssprache OPS5, 1987
This paper presents a novel real-time 3D-data acquisition system. The system that also provides a... more This paper presents a novel real-time 3D-data acquisition system. The system that also provides a normal 2D color image of the scene is based on the structured light approach. It is designed to combine conflicting requirements for 3D-data acquisition systems – a low overall system cost, dense as well as accurate range data and a minimization of the constraints imposed on the scene. Following an extensive review of the state of the art a description of the system is given. First results and the plausibility of the concept are evaluated.
Abstract. A lot of promising computer vision research has been con-ducted in order to automatical... more Abstract. A lot of promising computer vision research has been con-ducted in order to automatically recognize facial expressions during the last decade. Some of them achieve high accuracy, however, it has not yet been investigated how accurately humans accomplish this task, which will introduce a comparable measure. Therefore, we conducted a survey on this issue and this paper evaluates the gathered information regarding the recognition rate and the confusion of facial expressions. 1
Abstract—Models are often used to gain information about real-world objects. Their parameters des... more Abstract—Models are often used to gain information about real-world objects. Their parameters describe various proper-ties of the modeled object, such as position or deformation. In order to fit the model to a given image, displacement experts serve as an update function on the model parameterization. However, building robust displacement experts is a non-trivial task, especially in real-world environments. We propose a novel approach that learns displacement experts from a multi-band image representation which is specifically tuned towards the task of face model fitting. We provide the fitting algorithm not only the original image but an image representation that reflects the location of several facial components within the face. To demonstrate its capability to work robustly not only in constrained conditions, we integrate the Labeled Faces In The Wild database, which consists of images that have been taken outside lab or office environments. Our evaluation demonstrates, that the ...
This paper presents a complete system for automatic facial expression recognition. The Candide-3 ... more This paper presents a complete system for automatic facial expression recognition. The Candide-3 face model is used in conjunction with a learned objective function for face model fitting. The resulting sequence of model parameters is then presented to a recurrent neural network for classification. The advantage of using a recurrent network is that the temporal dependencies present in the image sequences can be taken into account during the classification. Since the entire process is automatic, and the recurrent networks can be used to make online predictions, the system would be ideal for real-time recognition. This would make it suitable for the CoTeSys ‘coffee break’ scenario, where guests must be recognised and served by robot waiters. Promising experimental results are presented on the Cohn-Kanade database.
Proceedings of the 10th International Symposium on Computer Science in Sports (ISCSS), 2015
While ball possession usually is considered on team level, a model on player level brings several... more While ball possession usually is considered on team level, a model on player level brings several advantages. We calculate ball possession and control statistics for all players as well as new ball control heat maps to evaluate the players’ performances. Furthermore, a basis for detecting events and tactical structure becomes available. To derive individual ball possession from spatio-temporal data, we present an automatic approach, based both on physical knowledge and machine learning techniques. Moreover, we introduce different ball possession definitions and algorithms to model various grades of ball control. When applied to flawless raw data, the algorithms show precision and recall ratios between 80 and 92 %. With approximately four percentage points less in uncorrected data, the presented algorithms are also reliable in real-world scenarios.
Computer Analysis of Images and Patterns, 2009
2010 IEEE International Conference on Image Processing, 2010
Real-Time Image and Video Processing 2010, 2010
2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, 2010
Lecture Notes in Computer Science, 2010
Die Wissensrepräsentationssprache OPS5, 1987
IEEE Transactions on Robotics and Automation, 2002
2010 Third International Conference on Advances in Computer-Human Interactions, 2010
Lecture Notes in Computer Science, 2009
Lecture Notes in Computer Science, 2009
Lecture Notes in Computer Science, 2009
Die Wissensrepräsentationssprache OPS5, 1987
Das ursprungliche in LISP implementierte OPS5 wird von einer LISP-Umgebung aus geladen. Es besteh... more Das ursprungliche in LISP implementierte OPS5 wird von einer LISP-Umgebung aus geladen. Es besteht weiterhin ein Zugriff auf das LISP-System. Dadurch wird einerseits der Aufruf externer (LISP-) Routinen etwas vereinfacht (die Unterstutzungsroutinen sind LISP-Funktionen), andererseits werden aber bei Laufzeitfehlern sowohl Fehlermeldungen des OPS5-Systems wie auch die des zugrundeliegenden LISP-Systems ausgegeben. Fur einen OPS5-Anwender sind daher Kenntnisse von LISP unumganglich.
Die Wissensrepräsentationssprache OPS5, 1987
Die Wissensrepräsentationssprache OPS5, 1987
This paper presents a novel real-time 3D-data acquisition system. The system that also provides a... more This paper presents a novel real-time 3D-data acquisition system. The system that also provides a normal 2D color image of the scene is based on the structured light approach. It is designed to combine conflicting requirements for 3D-data acquisition systems – a low overall system cost, dense as well as accurate range data and a minimization of the constraints imposed on the scene. Following an extensive review of the state of the art a description of the system is given. First results and the plausibility of the concept are evaluated.
Abstract. A lot of promising computer vision research has been con-ducted in order to automatical... more Abstract. A lot of promising computer vision research has been con-ducted in order to automatically recognize facial expressions during the last decade. Some of them achieve high accuracy, however, it has not yet been investigated how accurately humans accomplish this task, which will introduce a comparable measure. Therefore, we conducted a survey on this issue and this paper evaluates the gathered information regarding the recognition rate and the confusion of facial expressions. 1
Abstract—Models are often used to gain information about real-world objects. Their parameters des... more Abstract—Models are often used to gain information about real-world objects. Their parameters describe various proper-ties of the modeled object, such as position or deformation. In order to fit the model to a given image, displacement experts serve as an update function on the model parameterization. However, building robust displacement experts is a non-trivial task, especially in real-world environments. We propose a novel approach that learns displacement experts from a multi-band image representation which is specifically tuned towards the task of face model fitting. We provide the fitting algorithm not only the original image but an image representation that reflects the location of several facial components within the face. To demonstrate its capability to work robustly not only in constrained conditions, we integrate the Labeled Faces In The Wild database, which consists of images that have been taken outside lab or office environments. Our evaluation demonstrates, that the ...
This paper presents a complete system for automatic facial expression recognition. The Candide-3 ... more This paper presents a complete system for automatic facial expression recognition. The Candide-3 face model is used in conjunction with a learned objective function for face model fitting. The resulting sequence of model parameters is then presented to a recurrent neural network for classification. The advantage of using a recurrent network is that the temporal dependencies present in the image sequences can be taken into account during the classification. Since the entire process is automatic, and the recurrent networks can be used to make online predictions, the system would be ideal for real-time recognition. This would make it suitable for the CoTeSys ‘coffee break’ scenario, where guests must be recognised and served by robot waiters. Promising experimental results are presented on the Cohn-Kanade database.
Proceedings of the 10th International Symposium on Computer Science in Sports (ISCSS), 2015
While ball possession usually is considered on team level, a model on player level brings several... more While ball possession usually is considered on team level, a model on player level brings several advantages. We calculate ball possession and control statistics for all players as well as new ball control heat maps to evaluate the players’ performances. Furthermore, a basis for detecting events and tactical structure becomes available. To derive individual ball possession from spatio-temporal data, we present an automatic approach, based both on physical knowledge and machine learning techniques. Moreover, we introduce different ball possession definitions and algorithms to model various grades of ball control. When applied to flawless raw data, the algorithms show precision and recall ratios between 80 and 92 %. With approximately four percentage points less in uncorrected data, the presented algorithms are also reliable in real-world scenarios.
Computer Analysis of Images and Patterns, 2009
2010 IEEE International Conference on Image Processing, 2010
Real-Time Image and Video Processing 2010, 2010
2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, 2010
Lecture Notes in Computer Science, 2010
Die Wissensrepräsentationssprache OPS5, 1987
IEEE Transactions on Robotics and Automation, 2002
2010 Third International Conference on Advances in Computer-Human Interactions, 2010
Lecture Notes in Computer Science, 2009
Lecture Notes in Computer Science, 2009
Lecture Notes in Computer Science, 2009