Novelty detection in the machine learning context refers to identifying unknown/novel data, i.e., data which vary greatly from the ones that the system was trained with. This paper explores this technique as applied to acoustic surveillance of abnormal situations. The ultimate goal of the system is to help an authorized person towards taking the appropriate actions for preventing life/property loss. A wide variety of acoustic parameters is employed towards forming a multidomain feature vector, which captures diverse characteristics of the audio signals. Subsequently the feature coefficients are fed to three probabilistic novelty detection methodologies. Their performance is computed using two measures which take into account misdetections and false alarms. Out dataset was recorded under real-world conditions including three different locations where various types of normal and abnormal sound events were captured. A smart-home environment, an open public space, and an office corridor were used. The results indicate that probabilistic novelty detection can provide an accurate analysis of the audio scene to identify abnormal events.

The present study presents a practical methodology for automatic space monitoring based solely on the perceived acoustic information. We consider the case where atypical situations such as screams, explosions and gunshots take place in a metro station environment. Our approach is based on a two stage recognition schema, each one exploiting HMMs for approximating the density function of the corresponding sound class. The main objective is to detect abnormal events that take place in a noisy environment. A thorough evaluation procedure is carried out under different SNR conditions and we report high detection rates with respect to false alarm and miss probabilities rates.

There is a rich literature on automatic species identification of a specific target taxon as regards various vocalizing animals. Research usually is restricted to specific species – in most cases a single one. It is only very recently that the number of monitored species has started to increase for certain habitats involving birds. Automatic acoustic monitoring has not yet been proven to be generic enough to scale to other taxa and habitats than the ones described in the original research. Although attracting much attention, the acoustic monitoring procedure is neither well established yet nor universally adopted as a biodiversity monitoring tool. Recently, the multi-instance multi-label framework on bird vocalizations has been introduced to face the obstacle of simultaneously vocalizing birds of different species. We build on this framework to integrate novel, image-based heterogeneous features designed to capture different aspects of the spectrum. We applied our approach to a taxon-rich habitat that included 78 birds, 8 insect species and 1 amphibian. This dataset constituted the Multi-label Bird Species Classification Challenge-NIPS 2013 where the proposed approach achieved an average accuracy of 91.25% on unseen data.

Τhe concept of remote insect surveillance at large spatial scales for many serious insect pests of agricultural and medical importance has been introduced in a series of our papers. We augment typical, low-cost plastic traps for many insect pests with the necessary optoelectronic sensors to guard the entrance of the trap to detect, time-stamp, GPS tag, and—in relevant cases—identify the species of the incoming insect from their wingbeat. For every important crop pest, there are monitoring protocols to be followed to decide when to initiate a treatment procedure before a serious infestation occurs. Monitoring protocols are mainly based on specifically designed insect traps. Traditional insect monitoring suffers in that the scope of such monitoring: is curtailed by its cost, requires intensive labor, is time consuming, and an expert is often needed for sufficient accuracy which can sometimes raise safety issues for humans. These disadvantages reduce the extent to which manual insect monitoring is applied and therefore its accuracy, which finally results in significant crop loss due to damage caused by pests. With the term ‘surveillance’ we intend to push the monitoring idea to unprecedented levels of information extraction regarding the presence, time-stamping detection events, species identification, and population density of targeted insect pests. Insect counts, as well as environmental parameters that correlate with insects’ population development, are wirelessly transmitted to the central monitoring agency in real time and are visualized and streamed to statistical methods to assist enforcement of security control to insect pests. In this work, we emphasize how the traps can be self-organized in networks that collectively report data at local, regional, country, continental, and global scales using the emerging technology of the Internet of Things (IoT). This research is necessarily interdisciplinary and falls at the intersection of entomology, optoelectronic engineering, data-science, and crop science and encompasses the design and implementation of low-cost, low-power technology to help reduce the extent of quantitative and qualitative crop losses by many of the most significant agricultural pests. We argue that smart traps communicating through IoT to report in real-time the level of the pest population from the field straight to a human controlled agency can, in the very near future, have a profound impact on the decision-making process in crop protection and will be disruptive of existing manual practices. In the present study, three cases are investigated: monitoring Rhynchophorus ferrugineus (Olivier) (Coleoptera: Curculionidae) using (a) Picusan and (b) Lindgren trap; and (c) monitoring various stored grain beetle pests using the stored-grain pitfall trap. Our approach is very accurate, reaching 98–99% accuracy on automatic counts compared with real detected numbers of insects in each type of trap.

In this paper we use multiple microphone arrays to fuse the location estimate from each microphone array which yields an improved estimate of the positions and velocities of multiple, simultaneously active, moving speakers based on time delay of arrivals (TDOAs). Our approach: 1) incorporates kinematic information of moving speakers by using an interacting multiple model (IMM) estimator for each speaker in order to constrain the evolution of the location measurements; 2) fuses the location estimates of the same speaker from multiple microphone arrays for better acoustical coverage of the sensed environment, and 3) directly accounts for the measurement origin uncertainty, i.e., which measurement comes from which speaker by using the probabilistic data association (PDA) technique with the IMM estimator. We demonstrate that a network of arrays combined with the estimation technique widely used in multisensor multitarget tracking area provides a consistent and coherent way to reduce the uncertainty and ambiguity of measurements. The effectiveness of our approach is illustrated by extensive simulation study on tracking a single moving speaker and two closely-spaced speakers with a crossing segment in their trajectories.