In this paper, we present a comparative evaluation of various multi-person tracking-by-detection approaches on public data sets. This paper investigates five popular trackers coupled with six relevant visual people detectors evaluated on seven public data sets. The evaluation emphasizes on exhibited performance variation depending on tracker-detector choices. Our experimental results show that the overall performance depends on how challenging the data set is, the performance of the detector on the specific data set, and the tracker-detector combination. Some trackers are more sensitive to the choice of a detector and some detectors to the choice of a tracker than others. Based on our results, two of the trackers demonstrate the best performances consistently across different data sets, whereas the best performing detectors vary per data set. This underscores the need for careful application context specific evaluation when choosing a detector.

This paper presents a novel framework for learning a soft-cascade detector with explicit computation time considerations. Classically, training techniques for softcascade detectors select a set of weak classifiers and their respective thresholds, solely to achieve the desired detection performance without any regard to the detector response time. Nevertheless, since computation time performance is of utmost importance in many time-constrained applications, this work divulges an optimization approach that aims to minimize the mean cascade response time, given a desired detection performance, fixed beforehand. The resulting problem is NP-Hard, therefore finding an optimal threshold vector can be very time-consuming, especially when building a soft-cascade detector of long length. An efficient local search procedure is presented that deals with long-length detectors. Our evaluations on two challenging public datasets confirm that a faster cascade detector can be learned while maintaining similar detection performances.

Perceptual systems deployed in embedded intelligent systems are handicapped with limited computation resources. As a result, they require explicit computation time consideration during design time. In this paper, we investigate the impact a visual person detector explicitly optimized to minimize computational time has on an exemplar robotic application - navigation in human shared environments. The experimental results attest the improved perceptual frame rate leads to significant improvements in vision system performance and robotic mission success rate.

We present a multi-modal perception based architecture to realize a non-intrusive domestic assistant robot. The realized robot is non-intrusive in that it only starts interaction with a user when it detects the user's intention to do so automatically. All the robot's actions are based on multi-modal perceptions, which include: user detection based on RGB-D data, user's intention-for-interaction detection with RGB-D and audio data, and communication via speech recognition. The utilization of multi-modal cues in different parts of the robotic activity paves the way to successful robotic runs.

In this paper, the problem of minimizing the mean response-time of a soft-cascade detector is addressed. A soft-cascade detector is a machine learning tool used in applications that need to recognize the presence of certain types of object instances in images. Classical soft-cascade learning methods select the weak classifiers that compose the cascade, as well as the classification thresholds applied at each cascade level, so that a desired detection performance is reached. They usually do not take into account its mean response-time, which is also of importance in time-constrained applications. To overcome that, we consider the threshold selection problem aiming to minimize the computation time needed to detect a target object in an image (i.e., by classifying a set of samples). We prove the NP-hardness of the problem and propose a mathematical model that takes benefit from several dominance properties, which are put into evidence. On the basis of computational experiments, we show that we can provide a faster cascade detector, while maintaining the same detection performances.

In this paper, we investigate the notion of incorporating feature computation time (CT) measures during feature selection in a boosted cascade people detector utilizing heterogeneous pool of features. We present various approaches based on pareto-front analysis, CT weighted adaboost, and Binary Integer Programming (BIP) with comparative evaluations. The novel feature selection method proposed based on BIP — the main contribution — mines heterogeneous features taking both detection performance and CT explicitly into consideration. The results demonstrate that the detector using this feature selection scheme exhibits low miss rates (MRs) with significant boost in frame rate. For example, it achieves a [Formula: see text] less MR at [Formula: see text] FPPW compared to Dalal and Triggs HOG detector with a [Formula: see text]x speed improvement. The presented extensive experimental results clearly highlight the improvements the proposed framework brings to the table.

In this work, we present a comparative evaluation of various `tracking-by-detection' approaches on public datasets. The work investigates popular sequential Monte Carlo and template ensemble based trackers coupled with relevant visual people detectors with emphasis on exhibited performance variation depending on tracker-detector choice. Extensive experimental results are provided on public dataset and results indicate the choice of a detector can significantly vary the performance of a tracker. Our experimental results show, depending on the choice of the detector, average tracking accuracy across three public datasets could exhibit a 45% standard deviation with only, on average, a 6.8% and 11.1% standard deviation in detector recall and precision respectively.

Understanding people's intention, be it action or thought, plays a fundamental role in establishing coherent communication amongst people, especially in non-proactive robotics, where the robot has to understand explicitly when to start an interaction in a natural way. In this work, a novel approach is presented to detect people's intention-for-interaction. The proposed detector fuses multimodal cues, including estimated head pose, shoulder orientation and vocal activity detection, using a probabilistic discrete state Hidden Markov Model. The multimodal detector achieves up to 80% correct detection rates improving purely audio and RGB-D based variants.

In this paper we present a novel people detector that employs discrete optimization for feature selection. Specifically, we use binary integer programming to mine heterogeneous features taking both detection performance and computation time explicitly into consideration. The final trained detector exhibits low Miss Rates with significant boost in frame rate. For example, it achieves a 2.6% less Miss Rate at 10-4 FPPW compared to Dalal and Triggs HOG detector with a 9.22x speed improvement.

In this paper, we present a fast Histogram of Oriented Gradients (HOG) based person detector. The detector adopts a cascade of rejectors framework by selecting discriminant features via a new proposed feature selection framework based on Binary Integer Programming. The mathematical programming explicitly formulates an optimization problem to select discriminant features taking detection performance and computation time into account. The learning of the cascade classifier and its detection capability are validated using a proprietary dataset acquired using the Ladybug2 spherical camera and the public INRIA person detection dataset. The final detector achieves a comparable detection performance as Dalal and Triggs [2] detector while achieving on average more than 2.5×-8× speed up depending on the training dataset.

In this paper, a boosted cascade person detection framework with heterogeneous pool of features is presented. The framework unveils a new feature selection scheme based on Pareto-Front analysis and AdaBoost. At each cascade node, Pareto-Front analysis is used to select dominant features thereby reducing the total number of features to a size easily manageable by AdaBoost. The final detector achieves a very low Miss Rate of 0.07 at 10-4 False Positives Per Window on the INRIA public dataset.