Detection is an inherent part of every advanced automatic tracking system. In this work we focus on automatic detection of humans by enhanced background subtraction. Background subtraction (BS) refers to the process of segmenting moving regions from video sensor data and is usually performed at pixel level. In its standard form this technique involves building a model of the background and extracting regions of the foreground. In this paper, we propose a cluster-based BS technique using a mixture of Gaussians. An adaptive mechanism is developed that allows automated learning of the model parameters. The efficiency of the designed technique is demonstrated in comparison with a pixel-based BS.
The growing popularity of social media raises concerns about children's online safety. Of particular concern are interactions between minors and adults with predatory intentions. Unfortunately, previous research on online sexual grooming has relied on time-intensive manual annotation by domain experts, limiting both the scale and scope of possible interventions. This work explores the possibility of detecting predatory behaviours with accuracy comparable to expert annotators using machine learning (ML). Using a dataset of 6771 chat messages sent by child sex offenders, labelled by two of the authors who are forensic psychology experts, we study how well can deep learning algorithms identify eleven known predatory behaviours. We find that the best-performing ML models are consistent but not on par with expert annotation. We therefore consider a system where an expert annotator validates the ML algorithms outputs. The combination of human decision-making and computer efficiency yields precision—but not recall—comparable to manual annotation, while taking only a fraction of the time needed by a human annotator. Our findings underscore the promise of ML as a tool for assisting researchers in this area, but also highlight the current limitations in reliably detecting online sexual exploitation using ML.
We propose a fully Bayesian approach to wideband, or broadband, direction-of-arrival (DoA) estimation and signal detection. Unlike previous works in wideband DoA estimation and detection, where the signals were modeled in the time-frequency domain, we directly model the time-domain representation and treat the non-causal part of the source signal as latent variables. Furthermore, our Bayesian model allows for closed-form marginalization of the latent source signals by leveraging conjugacy. To further speed up computation, we exploit the sparse ``stripe matrix structure'' of the considered system, which stems from the circulant matrix representation of linear time-invariant (LTI) systems. This drastically reduces the time complexity of computing the likelihood from $\mathcal{O}(N^3 k^3)$ to $\mathcal{O}(N k^3)$, where $N$ is the number of samples received by the array and $k$ is the number of sources. These computational improvements allow for efficient posterior inference through reversible jump Markov chain Monte Carlo (RJMCMC). We use the non-reversible extension of RJMCMC (NRJMCMC), which often achieves lower autocorrelation and faster convergence than the conventional reversible variant. Detection, estimation, and reconstruction of the latent source signals can then all be performed in a fully Bayesian manner through the samples drawn using NRJMCMC. We evaluate the detection performance of the procedure by comparing against generalized likelihood ratio testing (GLRT) and information criteria.
This article develops models and algorithms for continuous-discrete multiple target filtering, in which the multi-target system is modelled in continuous time and measurements are available at discrete time steps. In order to do so, this paper first proposes a statistical model for multi-target appearance, dynamics and disappearance in continuous time, based on continuous time birth/death processes and stochastic differential equations. The multitarget state is observed at known time instants based on the standard measurement model, and the objective is to compute the distribution of the multi-target state at these time steps. For the Wiener velocity model, we derive a closed-form formula to obtain the best Gaussian Poisson point process fit to the birth density based on Kullback-Leibler minimisation. The resulting discretised model gives rise to the continuous-discrete Gaussian Poisson multi-Bernoulli mixture (PMBM) filter, the continuous-discrete Gaussian mixture probability hypothesis density (PHD) filter and the continuous-discrete Gaussian mixture cardinality PHD (CPHD) filter. The proposed filters are specially useful for multi-target estimation when the time interval between measurements is non-uniform.
Semi-Markov models are a generalisation of Markov models that explicitly model the state-dependent sojourn time distribution, the time for which the system remains in a given state. Markov models result in an exponentially distributed sojourn time, while semi-Markov models make it possible to define the distribution explicitly. Such models can be used to describe the behaviour of manoeuvring targets, and particle filtering can then facilitate tracking. An architecture is proposed that enables particle filters to be both robust and efficient when conducting joint tracking and classification. It is demonstrated that this approach can be used to classify targets on the basis of their manoeuvrability.
'%3e%3cpath fill='%23012169' d='M0 0v30h60V0z'/%3e%3cpath stroke='%23fff' stroke-width='6' d='m0 0 60 30m0-30L0 30'/%3e%3cpath stroke='%23C8102E' stroke-width='4' d='m0 0 60 30m0-30L0 30' clip-path='url(%23b)'/%3e%3cpath stroke='%23fff' stroke-width='10' d='M30 0v30M0 15h60'/%3e%3cpath stroke='%23C8102E' stroke-width='6' d='M30 0v30M0 15h60'/%3e%3c/g%3e%3c/svg%3e)
