Tommaso Gritti

Learning Discriminative LBP-Histogram Bins for Facial Expression Recognition

Local Binary Patterns (LBP) have been well exploited for facial image analysis recently. In the existing work, the LBP histograms are extracted from local facial regions, and used as a whole for the regional description. However, not all bins in the LBP histogram are necessary to be useful for facial representation. In this paper, we propose to learn discriminative LBP-Histogram (LBPH) bins for the task of facial expression recognition. Our experiments illustrate that the selected LBPH bins provide a compact and discriminative facial representation. We experimentally illustrate that it is necessary to consider multiscale LBP for representing faces, and most discriminative information is contained in uniform patterns. By adopting SVM with the selected multiscale LBPH bins, we obtain the best recognition performance of 93.1% on the Cohn-Kanade database.

Local features based facial expression recognition with face registration errors

In this paper, we extensively investigate local features based facial expression recognition with face registration errors, which has never been addressed before. Our contributions are three fold. Firstly, we propose and experimentally study the histogram of oriented gradients (HOG) descriptors for facial representation. Secondly, we present facial representations based on local binary patterns (LBP) and local ternary patterns (LTP) extracted from overlapping local regions. Thirdly, we quantitatively study the impact of face registration errors on facial expression recognition using different facial representations. Overall LBP with overlapping gives the best performance (92.9% recognition rate on the Cohn-Kanade database), while maintaining a compact feature vector and best robustness against face registration errors.

Background Subtraction under Sudden Illumination Changes

Robust background subtraction under sudden illuminationchanges is a challenging problem. In this paper, wepropose an approach to address this issue, which combinesthe Eigenbackground algorithm together with a statisticalillumination model. The rst algorithm is used to give arough reconstruction of the input frame, while the secondone improves the foreground segmentation. We introduce anonline spatial likelihood model by detecting reliable backgroundand foreground pixels. Experimental results illustratethat our approach achieves consistently higher accuracycompared to several state-of-the-art algorithms

Flower power

Flower Power is an interactive installation that creates immersive light atmospheres by capturing the colors of flowers and re-projecting them in the environment. A camera is suspended over a flower bed and its position depends on the positions of the visitors around the installation, which are detected using presence sensors. The image captured by the camera is rendered into the surrounding environment as a suggestive, saturated light scene. The participants can thus give the power to their favorite flower to set the overall light experience. In this paper we describe the motivation and the realization of the installation, and we discuss insights and reactions collected during the 2010 STRP Festival.

A framework for robust feature selection for real-time fashion style recommendation

In this paper, we present the Smart Mirror system for fashion recommendation. The system uses intelligent vision technology to recognize clothing styles and supports real-time fashion recommendation. An important design challenge is to achieve sufficiently high style recognition accuracy while simultaneously offering robustness to input variations occurring in practice. We propose a framework for the selection of features that offer robust performance by assessing various evaluation measures under realistic deviations of optimal input data. The process is applied to a variety of low level features for clothing style description, including color histograms, local binary pattern (LBP) features and histogram of oriented gradient (HOG) features. We conclude the paper with an illustration of our results for web camera data and with a number of recommendations on how to move forward towards automatic fashion style perception.

Practical study on real-time hand detection

In this paper we describe algorithms and image features that can be used to construct a real-time hand detector. We present our findings using the Histogram of Oriented Gradients (HOG) features in combination with two variations of the AdaBoost algorithm. First, we compare stump and tree weak classifier. Next, we investigate the influence of a large training database. Furthermore, we compare the performance of HOG against the Haar-like features.

Detecting customers' buying events on a real-life database

Video Analytics covers a large set of methodologies which aim at automatically extracting information from video material. In the context of retail, the possibility to effortlessly gather statistics on customer shopping behavior is very attractive. In this work, we focus on the task of automatic classification of customer behavior, with the objecting to recognize buying events. The experiments are performed on several hours of video collected in a supermarket. Given the vast effort of the research community on the task of tracking, we assume the existence of a video tracking system capable of producing a trajectory for every individual, and currently manually annotate the input videos with trajectories. From the annotated video recordings, we extract features related to the spatio-temporal behavior of the trajectory, and to the user movement, and analyze the shopping sequences using a Hidden Markov Model (HMM). First results show that it is possible to discriminate between buying and non-buying behavior with an accuracy of 74%.

ImagiLight: a vision approach to lighting scene setting

The advent of integrated lighting installations, consisting of individually controllable lamps with advanced rendering capabilities, is fundamentally transforming lighting. This brings a need for an intuitive control capable of fully exploiting the rendering capabilities of the complete lighting infrastructure. In this paper we present a new method to automatically create lighting atmospheres in any type of environment, that allows for an natural interaction with the lighting system and generates unique, suggestive effects. To prove the effectiveness and versatility of the proposed solution, we deploy the system in several application scenarios, and discuss the results.

Toward fully automated face pose estimation

In this paper we focus on the task of fully automatic real-time face 3D pose estimation, both person independent and calibration free. We developed a complete system, which is capable of self initializing, estimates the pose robustly and detects failure of tracking. As a first contribution, we describe the initialization step, which does not rely on any user interaction. As a second contribution we detail a robust tracking methodology, capable of dealing with fast user motion and varying lighting conditions. This includes improvement on both the matching error metric and the search algorithms. We show how the choice of the texture representation can strongly influence the stability of the pose estimation. We finally evaluate the performance of the system on realistic videos. The results show that the proposed method is both adaptable to different users and robust to lighting changes.

Active Vision for Goal-Oriented Humanoid Robot Walking

Complex visual tasks may be tackled with remarkably simple neural architectures generated by a co-evolutionary process of active vision and feature selection. This hypothesis has recently been tested in several robotic applications such as shape discrimination, car driving, indoor/outdoor navigation of a wheeled robot. Here we describe an experiment where this hypothesis is further examined in goal-oriented humanoid bipedal walking task. Hoap-2 humanoid robot equipped with a primitive vision system on its head is evolved while freely interacting with its environment. Unlike wheeled robots, bipedal walking robots are exposed to largely perturbed visual input caused by their own walking dynamics. We show that evolved robots are capable of coping with the dynamics and of accomplishing the task by means of active, efficient camera control.