Poor Lighting Conditions Research Articles

Optimal capacity allocation and scheduling strategy for CSP+PV hybrid standalone power plants

Solar photovoltaic (PV) power generation, as an important clean technology, has been widely adopted globally, especially in remote island areas where access to the main power grid is unavailable. PV power can serve as the primary energy source for standalone island grids. However, due to its dependence on sunlight, PV power output fluctuates, particularly during nighttime and under poor lighting conditions, necessitating the integration of energy storage technology or alternative power generation methods to ensure continuous power supply. Concentrating solar power (CSP) generation, as an emerging technology, can provide efficient power output when solar radiation is abundant and ensure continuous power supply through thermal energy storage systems during adverse weather conditions or nighttime. Although CSP offers advantages such as dispatchability, its high construction and maintenance costs may pose challenges in commercial deployment. Hybrid solar power plants combining both PV and CSP technologies leverage the strengths of both, ensuring more stable and economically viable power output. This study establishes a model for hybrid solar power plants, considering the impact of PV and CSP component capacities and proportions on performance and costs, i.e., capacity allocation. To maximize the overall benefits of standalone microgrids while ensuring the stability of the power station, a capacity allocation method guided by economic dispatch is proposed. Through iterative analysis, the optimal configuration is determined to minimize the system's equivalent annual costs. Simulation experiments validate the financial feasibility of hybrid solar power plants and the reliability of the proposed configuration method.

RRBM-YOLO: Research on Efficient and Lightweight Convolutional Neural Networks for Underground Coal Gangue Identification.

Coal gangue identification is the primary step in coal flow initial screening, which mainly faces problems such as low identification efficiency, complex algorithms, and high hardware requirements. In response to the above, this article proposes a new "hardware friendly" coal gangue image recognition algorithm, RRBM-YOLO, which is combined with dark light enhancement. Specifically, coal gangue image samples were customized in two scenarios: normal lighting and simulated underground lighting with poor lighting conditions. The images were preprocessed using the dim light enhancement algorithm Retinexformer, with YOLOv8 as the backbone network. The lightweight module RepGhost, the repeated weighted bi-directional feature extraction module BiFPN, and the multi-dimensional attention mechanism MCA were integrated, and different datasets were replaced to enhance the adaptability of the model and improve its generalization ability. The findings from the experiment indicate that the precision of the proposed model is as high as 0.988, the mAP@0.5(%) value and mAP@0.5:0.95(%) values increased by 10.49% and 36.62% compared to the original YOLOv8 model, and the inference speed reached 8.1GFLOPS. This indicates that RRBM-YOLO can attain an optimal equilibrium between detection precision and inference velocity, with excellent accuracy, robustness, and industrial application potential.

NeRF-based Localization and Meshing with Wearable Laser Scanning System: A Case Study in Underground Environment

Abstract. Due to the subterranean scene’s poor lighting conditions and variability of the environment, real-time localizing and meshing in complex underground scenes present a challenging task, with high-potential applications in mining and tunnel protection. In this work, we propose a method that combines SLAM and NeRF for mesh reconstruction in the underground environment with a wearable device. First, a LiDAR-Inertial odometry is used for pose estimation. The resulting poses and sequential laser frames are synchronized to generate a novel data structure, scan-block, which is crucial for improving efficiency and ensuring local precision. This structure is designed to enhance efficiency and ensure local precision. Finally, the generated scan-blocks are passed to the backend NeRF for real-time mesh reconstruction. The experiments are carried out in a complex underground space. The experimental results proved that this method achieved good results. The demo video can be found at https://youtu.be/_y1vbaW06EM?si= hFnAl12u-ffU933h.

Adaptive positioning of large-scale aircraft components using a simplified image-based visual servoing via online measurement of ball-head center

In digital alignment systems, ensuring fast and accurate alignment between the ball head on large aircraft components and the ball socket of the positioner is the key to aircraft assembly efficiency and quality. Image-based visual servoing is one of the most advanced techniques with high efficiency and positioning accuracy, but it does not account for measuring the ball-head position. Moreover, variable target positions, similar feature interference, and poor lighting conditions pose great challenges to the fast and robust visual measurement of ball-head position. Hence, an online visual measurement method is introduced to measure the absolute position of the ball head in real time. An adaptive region of interest extraction algorithm and an improved edge following method are introduced to extract the elliptical contour from the ball-head image, and ellipse detection is achieved through a simple triplet selection algorithm. A geometric model based on the perspective projection of the sphere is established for estimating the ball-head center. Then, the simplified image-based visual servoing method is developed by constantizing the image Jacobian matrix. Next, an adaptive positioning method of large-scale aircraft components with multiple ball heads is proposed from a holistic perspective. Finally, experiments show that the final positioning accuracy of each ball head in the X/Y plane is less than ± 0.05 mm, with an efficiency almost ten times that of the conventional method. After adaptive adjustment, the position and angle errors of the simulated component are reduced within 0.6 mm and 0.3°, respectively.

Optimizing Classroom Lighting: Balancing Illuminance Uniformity and Glare Control with Led Technology

Indoor lighting environment significantly impacts students’ concentration and learning performance. Optimal lighting conditions enhance reading speed, accuracy, and comprehension abilities. In contrast, poor lighting conditions may lead to decreased concentration and increased errors in tasks. Therefore, classroom lighting is an important factor to consider in classroom design to optimize cognitive efficiency of students. Current classroom lighting solutions using LED systems have met European Norm (EN) standard in terms of illuminance and energy efficiency. However, although the issue of glare has been addressed, it has not been completely resolved. This study introduces a solution for glare-free classroom lighting that maintains uniformity while adhering to existing standards. This approach aims to create a visually comfortable learning environment, balancing effective lighting with glare reduction.

Rapid SLAM Method for Star Surface Rover in Unstructured Space Environments

The space environment is characterized by unstructured features, sparsity, and poor lighting conditions. The difficulty in extracting features from the visual frontend of traditional SLAM methods results in poor localization and time-consuming issues. This paper proposes a rapid and real-time localization and mapping method for star chart surveyors in unstructured space environments. Improved localization is achieved using multiple sensor fusion to sense the space environment. We replaced the traditional feature extraction module with an enhanced SuperPoint feature extraction network to tackle the challenge of challenging feature extraction in unstructured space environments. By dynamically adjusting detection thresholds, we achieved uniform detection and description of image keypoints, ultimately resulting in robust and accurate feature association information. Furthermore, we minimized redundant information to achieve precise positioning with high efficiency and low power consumption. We established a star surface rover simulation system and created simulated environments resembling Mars and the lunar surface. Compared to the LVI-SAM system, our method achieved a 20% improvement in localization accuracy for lunar scenarios. In Mars scenarios, our method achieved a positioning accuracy of 0.716 m and reduced runtime by 18.682 s for the same tasks. Our approach exhibits higher localization accuracy and lower power consumption in unstructured space environments.

SfM Photogrammetry for Cost-Effective 3D Documentation and Rock Art Analysis of the Dombate Dolmen (Spain) and the Megalithic Sites of Chã dos Cabanos and Chã da Escusalha (Portugal)

SfM (structure from motion) photogrammetry is a technique developed in the field of computer vision that enables the generation of three-dimensional (3D) models from a set of overlapping images captured from disparate angles. The application of this technique in the field of cultural heritage, particularly in the context of megalithic monuments, is inherently challenging due to the spatial constraints of these environments and the usual limitations posed by their architectural design, which often results in poor lighting conditions. This article presents an accurate and cost-efficient methodology for the study and documentation of rock art, which has been applied to three megalithic monuments in the Iberian Peninsula: one in Spain and two in Portugal. The three working environments are complex, but the combination of techniques used and improvements such as rendering for the enhancement of engravings and the creation of 3D stop-motion models made it possible to integrate all the information in 3D formats that allow its universal dissemination. This not only preserves the heritage in graphic form but also makes it accessible to the public, both for study and for virtual visits.

Seeing is believing: An Augmented Reality application for Palaeolithic rock art

By developing new recording methodologies, current rock art studies generate a large amount of graphic information about sites (tracings, photographs, three-dimensional reproductions) providing visibility of this fragile and little-known heritage, whose accessibility is often difficult or impossible for the general public. In addition, many rock art depictions are challenging to observe, due to the very nature of the artistic entities (fine engravings or faded paintings in karst environments or open-air sites with poor or changing light conditions), or to conservation problems derived from natural factors such as erosion and geological and biological processes, as well as from anthropic factors. These conditions make rock art depictions nearly indistinguishable in many places and on many objects today, except for experts. This difficulty of accessing and visualising rock art heritage, located in fragile environments and often challenging places such as caves or difficult-to-reach open-air sites, makes the information and knowledge generated by investigation of this heritage asset difficult to transfer to society in general, which is frequently unaware of the priceless value of this heritage. The present study proposes generating several mechanisms to transfer the results of research, restitution and documentation of rock art to society in general. An AR (Augmented Reality) application has been developed using LiDAR (Light Detection and Ranging) technology to address current challenges in implementing AR technologies in low-light environments. So far, this app has been developed in a Proof-of-Concept project at Spanish archaeological sites such as Hornos de la Peña (Cantabria), Domingo García (Segovia) and La Salud (Salamanca). This application will be particularly interesting for sites currently visited with or without a guide, allowing user interactivity and real-time reconstruction, for example, of the visibility of graphic motifs.

Implementation and Efficient Analysis of Preprocessing Techniques in Deep Learning for Image Classification

Deep learning models have recently been preferred to perform certain image-processing tasks. Recently, with the increasing radiation, heat, and poor lighting conditions, the raw image samples may contain noisy and ambiguous information. To process these images, the deep learning model requires a large number of data samples to learn the missing information from other clear data samples. This necessitates training the neural network with a huge dataset. The researchers are now attempting to filter and improve such noisy images via preprocessing in order to provide valid and accurate feature information to the neural network layers. However, certain research studies claim that some useful information may be lost when the image is not preprocessed with an appropriate filter or enhancement technique. The MSA (meta-synthesis and analysis) approach is utilized in this work to present the impact of the image processing applications done with and without preprocessing steps. Also, this work summarizes the existing deep learning-based image processing models utilizing or not preprocessing steps in their implementation. This work has also found that 85% of the existing techniques involve a preprocessing step while developing a deep learning model. However, a maximum accuracy of 96.89% is observed on Sine-Net when it is implemented without a preprocessing and the same model gave 96.85% when implemented with preprocessing. This research provides various research insights on the requirement and non-requirement of preprocessing steps in a deep learning-based implementation.

Automated extrinsic calibration of solid-state frame LiDAR sensors with non-overlapping field of view for monitoring indoor stockpile storage facilities

Several industrial and commercial bulk material management applications rely on accurate, current stockpile volume estimation. Proximal imaging and LiDAR sensing modalities can be used to derive stockpile volume estimates in outdoor and indoor storage facilities. Among available imaging and LiDAR sensing modalities, the latter is more advantageous for indoor storage facilities due to its ability to capture scans under poor lighting conditions. Evaluating volumes from such sensing modalities requires the pose (i.e., position and orientation) parameters of the used sensors relative to a common reference frame. For outdoor facilities, a Global Navigation Satellite System (GNSS) combined with an Inertial Navigation System (INS) can be used to derive the sensors’ pose relative to a global reference frame. For indoor facilities, GNSS signal outages will not allow for such capability. Prior research has developed strategies for establishing the sensor position and orientation for stockpile volume estimation while relying on multi-beam spinning LiDAR units. These approaches are feasible due to the large range and Field of View (FOV) of such systems that can capture the internal surfaces of indoor storage facilities.The mechanical movement of multi-beam spinning LiDAR units together with the harsh conditions within indoor facilities (e.g., excessive humidity, wide range of temperature variation, dust, and corrosive environment in deicing salt storage facilities) limit the use of such systems. With the increasing availability of solid-state LiDAR units, there is an interest in exploring their potential for stockpile volume estimation. Despite their higher robustness to harsh conditions, solid-state LiDAR units have shorter distance measurement range and limited FOV when compared with multi-beam spinning LiDAR. This research presents a strategy for the extrinsic calibration (i.e., estimating the relative pose parameters) of installed solid-state LiDAR units inside stockpile storage facilities. The extrinsic calibration is made possible using deployed spherical targets and a complete, reference scan of the facility from another LiDAR sensing modality. The proposed research introduces strategies for: 1) automated extraction of the spherical targets; 2) automated matching of these targets in the solid-state LiDAR and reference scans using invariant relationships among them; and 3) coarse-to-fine estimation of the calibration parameters. Experimental results in several facilities have shown the feasibility of using the proposed methodology to conduct the extrinsic calibration and volume evaluation with an error percentage less than 3.5% even with occlusion percentages reaching up to 50%.

Machine learning-based understanding of aquatic animal behaviour in high-turbidity waters

Inspired by the ambitions envisioned in the Fourth Industrial Revolution for aquaculture, also known as Aquaculture 4.0, the aquaculture (marine animal farming) industry is seeking to adopt data-driven Artificial Intelligence (AI) to help significantly improve business operations. One of the major barriers is the manual annotation of animal behaviour data, which is a time-consuming task that demands high levels of concentration from biologists. To address this challenge, this paper proposes novel automatic animal behaviour monitoring tailored for industrial scenarios. Our approach introduces a real-time machine-learning-based instance segmentation system that is specialised for underwater environments, where large groups of shrimp are farmed. The implemented system achieves an accuracy rate of 89% at 30 frames per second (fps) and can accurately detect shrimp in high-density areas under poor lighting conditions and high turbidity waters, despite the challenges of occlusion and overlapping. A key innovation of our method is the implementation of a new density cluster algorithm for time series and video analysis. This approach provides a more efficient and accurate way of monitoring animal behaviour, significantly saving time and effort for biologists and advancing the capabilities of automated aquaculture systems.

HR-YOLO: A Multi-Branch Network Model for Helmet Detection Combined with High-Resolution Network and YOLOv5

Automatic detection of safety helmet wearing is significant in ensuring safe production. However, the accuracy of safety helmet detection can be challenged by various factors, such as complex environments, poor lighting conditions and small-sized targets. This paper presents a novel and efficient deep learning framework named High-Resolution You Only Look Once (HR-YOLO) for safety helmet wearing detection. The proposed framework synthesizes safety helmet wearing information from the features of helmet objects and human pose. HR-YOLO can use features from two branches to make the bounding box of suppression predictions more accurate for small targets. Then, to further improve the iterative efficiency and accuracy of the model, we design an optimized residual network structure by using Optimized Powered Stochastic Gradient Descent (OP-SGD). Moreover, a Laplace-Aware Attention Model (LAAM) is designed to make the YOLOv5 decoder pay more attention to the feature information from human pose and suppress interference from irrelevant features, which enhances network representation. Finally, non-maximum suppression voting (PA-NMS voting) is proposed to improve detection accuracy for occluded targets, using pose information to constrain the confidence of bounding boxes and select optimal bounding boxes through a modified voting process. Experimental results demonstrate that the presented safety helmet detection network outperforms other approaches and has practical value in application scenarios. Compared with the other algorithms, the proposed algorithm improves the precision, recall and mAP by 7.27%, 5.46% and 7.3%, on average, respectively.

Implementing All-Weather Photocatalysis of Exhaust Fumes Based on the g-C3N4/TiO2/SrAl2O4: Eu2+, Dy3+ Ternary Composite Coating

This study examines the use of SrAl2O4: Eu2+, Dy3+ long-afterglow materials doped into g-C3N4/TiO2 coatings for photodegradation. The prepared sample was tested for the purification of automotive exhaust fumes, with the optimal mass ratio of g-C3N4/TiO2 and SrAl2O4: Eu2+, Dy3+ determined to be 1:1. Characterization tests, including XRD, FT-IR, XPS, and TG-DSC, were conducted to evaluate the microstructure and properties of the samples. Under poor lighting conditions, g-C3N4/TiO2 reduced CH and NOx by 59 ppm and 13 ppm within 4 h, respectively, while g-C3N4/TiO2/SrAl2O4: Eu2+, Dy3+ decreased CH and NOx by 98ppm and 34ppm, respectively, resulting in a significant improvement in degradation efficiency. The addition of long-afterglow materials significantly improves the efficiency of photocatalysts in purifying exhaust fumes in low-light environments, providing potential value for all-weather exhaust treatment in the future.

A lightweight network model designed for alligator gar detection

When using advanced detection algorithms to monitor alligator gar in real-time in wild waters, the efficiency of existing detection algorithms is subject to certain limitations due to turbid water quality, poor underwater lighting conditions, and obstruction by other objects. In order to solve this problem, we developed a lightweight real-time detection network model called ARD-Net, from the perspective of reducing the amount of calculation and obtaining more feature map patterns. We introduced a cross-domain grid matching strategy to accelerate network convergence, and combined the involution operator and dual-channel attention mechanism to build a more lightweight feature extractor and multi-scale detection reasoning network module to enhance the network’s response to different semantics. Compared with the yoloV5 baseline model, our method performs equivalently in terms of detection accuracy, but the model is smaller, the detection speed is increased by 1.48 times, When compared with the latest State-of-the-Art (SOTA) method, YOLOv8, our method demonstrates clear advantages in both detection efficiency and model size,and has good real-time performance. Additionally, we created a dataset of alligator gar images for training.

Collaborative brightening and amplification of low-light imagery via bi-level adversarial learning

Poor light conditions constrain the high pursuit of clarity and visible quality of photography especially smartphone devices. Admittedly, existing specific image processing methods, whether super-resolution methods or low-light enhancement methods, can hardly simultaneously enhance the resolution and brightness of low-light images at the same time. This paper dedicates a specialized enhancer with a dual-path modulated-interactive structure to recover high-quality sharp images in conditions of near absence of light, dubbed CollaBA, which learns the direct mapping from low-resolution dark-light images to their high-resolution normal sharp version. Specifically, we construct the generative modulation prior, serving as illuminance attention information, to regulate the exposure level of the neighborhood range. In addition, we construct an interactive degradation removal branch that progressively embeds the generated intrinsic prior to recover high-frequency detail and contrast at the feature level. We also introduce a multi-substrate up-scaler to integrate multi-scale sampling features, effectively addressing artifact-related problems. Rather than adopting the naive time-consuming learning strategy, we design a novel bi-level implicit adversarial learning mechanism as our fast training strategy. Extensive experiments on benchmark datasets — demonstrate our model’s wide-ranging applicability in various ultra-low-light scenarios, across 8 key performance metrics with significant improvements, notably achieving a 35.8% improvement in LPIPS and a 23.1% increase in RMSE. The code will be available at https://github.com/moriyaya/CollaBA.

Deep learning-based three-dimensional crack damage detection method using point clouds without color information

Automated high-precision crack detection on building structures under poor lighting conditions poses a significant challenge for traditional image-based methods. Overcoming this challenge is crucial to enhance the practical applicability of structural health monitoring and rapid damage assessment, especially in post-disaster scenarios like earthquakes. To address this challenge, this paper presents a deep learning-based three-dimensional crack detection method that utilizes light detection and ranging (LiDAR) point cloud data. Our method is specifically designed to address crack detection without relying on color information input, resulting in high-precision and robust apparent damage detection. The key contribution of this paper is the NL-3DCrack model, which enables automated three-dimensional crack semantic segmentation. This model comprises a feature embedding module, an incomplete neighbor feature extraction module, a decoder, and morphological filtering. Notably, we introduce an innovative incomplete neighbor mechanism to effectively mitigate the impact of outliers. To validate the effectiveness of our proposed method, we establish two three-dimensional crack detection datasets, namely the Luding dataset and the terrestrial laser scanner dataset, which are based on earthquake disasters. Experimental results demonstrate that our method achieves remarkable performance, with an intersection-over-union of 39.62% and 51.33% on the respective test sets, surpassing existing point cloud-based semantic segmentation models. Ablation experiments further confirm the effectiveness of our approach. In summary, our method showcases exceptional crack detection performance on LiDAR data using only XYZI channels. With its high precision and reliable results, it offers significant utility in real-world applications, contributing to improved structural health monitoring and rapid damage assessment after disasters, particularly in post-earthquake scenarios.

Lightweight underwater image adaptive enhancement based on zero-reference parameter estimation network

Underwater images suffer from severe color attenuation and contrast reduction due to the poor and complex lighting conditions in the water. Most mainstream methods employing deep learning typically require extensive underwater paired training data, resulting in complex network structures, long training time, and high computational cost. To address this issue, a novel ZeroReference Parameter Estimation Network (Zero-UAE) model is proposed in this paper for the adaptive enhancement of underwater images. Based on the principle of light attenuation curves, an underwater adaptive curve model is designed to eliminate uneven underwater illumination and color bias. A lightweight parameter estimation network is designed to estimate dynamic parameters of underwater adaptive curve models. A tailored set of non-reference loss functions are developed for underwater scenarios to fine-tune underwater images, enhancing the network’s generalization capabilities. These functions implicitly control the learning preferences of the network and effectively solve the problems of color bias and uneven illumination in underwater images without additional datasets. The proposed method examined on three widely used real-world underwater image enhancement datasets. Experimental results demonstrate that our method performs adaptive enhancement on underwater images. Meanwhile, the proposed method yields competitive performance compared with state-of-the-art other methods. Moreover, the Zero-UAE model requires only 17K parameters, minimizing the hardware requirements for underwater detection tasks. What’more, the adaptive enhancement capability of the Zero-UAE model offers a new solution for processing images under extreme underwater conditions, thus contributing to the advancement of underwater autonomous monitoring and ocean exploration technologies.

THE RELATIONSHIP BETWEEN SANITARY HYGIENE AND PHYSICAL ENVIRONMENTAL CONDITIONS WITH COCKROACH DENSITY IN THE CANTEEN OF SRIWIJAYA UNIVERSITY INDRALAYA CAMPUS IN 2024

The canteen is one of the strategic places in the spread of diseases carried by vectors and disease-carrying animals such as cockroaches. The spread of disease in the canteen is also influenced by environmental sanitation hygiene. A dirty and humid environment is a good breeding ground for cockroaches. The government has established a policy to reduce the vector population to minimize the risk of impact due to the disease it causes. Research to determine the relationship between sanitary hygiene and physical environmental conditions with cockroach density in the canteen of Sriwijaya University Indralaya campus was conducted. The study was an observational analytic study using a cross sectional study design. A total of 50 canteens on campus were used as samples and were total sampling. Data on sanitary hygiene standards were collected from observational sheets and measurements of physical environmental conditions were made using a hygro thermometer, while cockroach density was measured based on the sticky trap method. Data were analyzed univariately and bivariately using the Chi-Square test. The results showed that of the 50 canteens sampled, 6 canteens were classified as canteens with sanitary hygiene that did not meet the Permenkes requirements. This is indicated by the high cockroach density with a p-value of 0.000 <0.05. This means that sanitary hygiene significantly affects cockroach density. It was concluded that the most cockroaches were found in canteens with poor lighting conditions, low room temperature, and high room humidity. Sequential p-values of 0.000; 0.002; 0.002 for the 3 observed variables indicate a poor environment affects cockroach breeding, while a p-value <0.05 is interpreted as a significant relationship between physical environmental conditions and cockroach density.

Application of Makeup Image Optimization Recommendation System through the Analysis of BeautyGAN Based on Deep Learning

The purpose of this study was to identify the makeup preference of users and suggest a method to optimize the makeup style by using the preferred image for each age group through the analysis of BeautyGAN. Through this, you can propose a customized makeup style that suits you, and provide beneficial services to the makeup industry and consumers. In addition, by developing and validating new methods that effectively combine deep learning and vision systems, we aim to innovate makeup-related image conversion technology and contribute to academic and practical advances in this field. For this purpose, reference images suitable for each image were collected to implement image optimization for each age group, the input data reflected the researcher’s image, and the face was aligned and resized, after removing images with low resolution or poor lighting conditions. As a result of the performance evaluation of the BeautyGAN model, it was confirmed that the existing image was 51.26%, which is close to the BeautyGAN image of 38.89%. These results are judged to be able to provide customized makeup style suggestions or adjusted makeup effects that reflect the user’s preferences from an academic point of view, and from a practical point of view, it will be possible to improve the quality of customized beauty services by suggesting makeup styles that suit the characteristics of customers more accurately and quickly.

CARgram: CNN-based accident recognition from road sounds through intensity-projected spectrogram analysis

Road surveillance systems play an important role in traffic monitoring and detecting hazardous events. In recent years, several artificial intelligence-based approaches have been proposed for this purpose, typically based on the analysis of the acquired video streams. However, occlusions, poor lighting conditions, and heterogeneity of the events may often reduce their effectiveness and reliability. To overcome the limitations mentioned, scientific and industrial research has therefore focused on integrating such solutions with audio recognition methods. By automatically identifying anomalous traffic sounds, e.g., car crashes and skids, they help reduce false positives and missed alarms. Following this trend, in this work, we propose an innovative pipeline for the analysis of intensity-projected audio spectrograms from streams of traffic sounds, which exploits both (i) a visual approach based on a custom, special-purpose Convolutional Neural Network for the identification of anomalous events on the sound signal; and, (ii) a novel multi-representational encoding of the input, which proved to significantly improve the recognition accuracy of the neural models. The validation results of the proposed pipeline on the public MIVIA dataset, with a 0.96% of false positive rate, showed to be the best performance against the state-of-the-art competitors. Notably, following such findings, a prototype implementation has been deployed on a real-world video surveillance infrastructure.