Here you find a selection of my academic publications.

Bringing Sketch Recognition into Your Hands

Nataneli, G., Faloutsos, P.

Computer Graphics and Applications, October 2011

Sketching is an important enabling technology for modern animation interfaces. We present an approach for the analysis of sketches that is flexible and works consistently across a variety of different software and hardware platforms. Importantly, our method suits well the restricting requirements of mobile devices. The proposed framework builds a semantic representation of informal drawings and uses this model along with other properties of the sketch to drive an output module. Our method is robust and can run in real-time even on devices with limited hardware resources. We fully implemented the proposed recognition pipeline on a Nintendo DS, an iPhone, and a regular PC. We showcase an application of this technology for driving facial expressions.


Automated Visual Analysis to Support Training in Laparoscopic Surgery

Nataneli, G., Kasper, F., Allen, B., Dutson, E., Faloutsos, P.

Symposium on the Science of Learning in Medical Education, October 2010

It is a major challenge to train surgeons with the motor skills required to perform laparoscopic surgery. To address this issue the Society of American Gastrointestinal and Endoscopic Surgeons (SAGES) adopted the Fundamentals of Laparoscopic Surgery (FLS) as a standardized tool set for certification and assessment of medical trainees. However, the process of certification currently requires the expensive time commitment of expert medical professionals and trainees lack the tools to objectively assess their own skill level. We tackle these problems by introducing a computer vision system that can fully analyze and track the accredited FLS Peg Transfer task from a single video stream, producing an exact quantification of all the relevant events of the performance. Our solution is fully automated, works with the single camera provided in the FLS Box Trainer equipment, and does not require any modification of the available tools. We validate our method by comparing our automatic results with a selection of manually annotated videos of the FLS task performed by novice and expert surgeons. Our results show that automated solutions such as ours can greatly improve the current state of medical training in Laparoscopic Surgery.

 


Visual Tracking of Laparoscopic Instruments in Standard Training Environments

Allen, B., Kasper, F., Nataneli, G., Dutson, E., Faloutsos, P.

Medicine Meets Virtual Reality, September 2010

We propose a method for accurately tracking the spatial motion of standard laparoscopic
instruments from video. By exploiting the geometric and photometric invariants common
to standard FLS training boxes, we achieve robust and accurate tracking of instruments
from unmodified video feeds of training tasks. The proposed method requires no
modifications to the FLS training box, camera or instruments, and provides robust and
accurate 2D tracking of instrument motions, which is then extended to full 3D tracking
with a geometric construction.


Automatic Method for Vision-Based Tracking of Laparoscopic Instruments

Allen, B., Kasper, F., Nataneli, G., Dutson, E., Faloutsos, P.

UCLA Technical Report No. 100027, September 2010

We propose a method for accurately tracking the spatial motion of standard laparoscopic
instruments from video. By exploiting the geometric and photometric invariants common
to standard FLS training boxes, we achieve robust and accurate tracking of instruments
from unmodified video feeds of training tasks. The proposed method requires no
modifications to the FLS training box, camera or instruments, and provides robust and
accurate 2D tracking of instrument motions, which is then extended to full 3D tracking
with a geometric construction.


Robust Classification of Strokes with SVM and Grouping

Nataneli, G., Faloutsos, P.

International Symposium on Visual Computing, November 2007

The ability to recognize the strokes drawn by the user is central to most sketch-based interfaces. However, very few solutions that rely on recognition are robust enough to make sketching a definitive alternative to traditional WIMP user interfaces. In this paper, we propose an approach based on classification that given an unconstrained sketch, can robustly assign a label to each stroke that comprises the sketch. A key contribution of our approach is a technique for grouping strokes that eliminates outliers and enhances the robustness of the classification. We also propose a set of features that capture important attributes of the shape and mutual relationship of strokes. These features are statistically well-behaved and enable robust classification with Support Vector Machines (SVM). We conclude by presenting a concrete implementation of these techniques in an interface for driving facial expressions.


Robust Classification of Strokes with SVM and Grouping

Nataneli, G., Faloutsos, P.

Lecture Notes in Computer Science, November 2007

The ability to recognize the strokes drawn by the user is central to most sketch-based interfaces. However, very few solutions that rely on recognition are robust enough to make sketching a definitive alternative to traditional WIMP user interfaces. In this paper, we propose an approach based on classification that given an unconstrained sketch, can robustly assign a label to each stroke that comprises the sketch. A key contribution of our approach is a technique for grouping strokes that eliminates outliers and enhances the robustness of the classification. We also propose a set of features that capture important attributes of the shape and mutual relationship of strokes. These features are statistically well-behaved and enable robust classification with Support Vector Machines (SVM). We conclude by presenting a concrete implementation of these techniques in an interface for driving facial expressions.


Sketching Facial Expressions

Nataneli, G., Faloutsos, P.

SIGGRAPH, Technical Sketches, August 2007

We present an innovative sketch-based interface for driving facial expressions. Unlike existing solutions our approach relies on recognition and constructs a semantically relevant representation of a sketched face. This representation is parameterized and used to drive a facial model. The main appeal of our method is that the interface is completely decoupled from the underlying facial model that is used. Therefore one single interface is capable of driving a variety of different models both 2D and 3D. The connection between our tool and the face model is defined by a library of template strokes that can be generated with ease


Sketch-Based Facial Animation

Nataneli, G., Faloutsos, P.

Symposium of Computer Animation, Posters and Demos, August 2006

Sketching is a powerful tool used by animators and graphic artists to capture their visions. Computer tools, on the other hand, are often at odds with such artistic vision and require artists to manually tweak numerous attributes in order to achieve desired results. Facial animation exposes this limitation more than other applications; typically a large number of parameters are needed to provide users with enough flexibility to produce convincing results. We propose a method that simplifies facial animation by enabling users to control facial expressions directly by means of sketches.


Plant Phenology - Terrestrial Ecology Observing Systems at the James Reserve

Borgstrom, J., Estrin, D., Graham, E., Hamilton, M., Kaiser, W., Nataneli, G., Rundel, P., Taggart, M., Wimbrow, M., Yuen, E.

CENS Poster Session, July 2005

We researched a semi-automated image-based method that can use the NIMS system installed at the James Reserve to estimate the area of leaves and the density of leaves in a specified region.


An Area-Optimality Study of Floorplanning

Cong, J., Nataneli, G., Romesis, M., Shinnerl, J.

International Symposium on Physical Design, March 2004

A novel algorithm for rectangular floorplanning with guaranteed 100% area utilization is  used to construct new sets of floorplanning benchmarks. By minimizing the maximum block aspect ratio subject to a zero-dead-space constraint, example zero-dead-space (ZDS) floorplans matching the area profiles of any existing floorplanning benchmark circuits can be constructed. A mathematical analysis shows that the aspect ratios of the ZDS benchmarks’ blocks are uniformly bounded within  in most cases. Block packings produced by the Parquet, B*-tree, TCG-S, and BloBB packages on these new benchmarks are compared to the optimal-area floorplans produced by the ZDS algorithm.