Deep Multimodal Detection in Reduced Visibility Using Thermal Depth Estimation for Autonomous Driving.

Recently, the rapid development of convolutional neural networks (CNN) has consistently improved object detection performance using CNN and has naturally been implemented in autonomous driving due to its operational potential in real-time. Detecting moving targets to realize autonomous driving is an essential task for the safety of drivers and pedestrians, and CNN-based moving target detectors have shown stable performance in fair weather. However, there is a considerable drop in detection performance during poor weather conditions like hazy or foggy situations due to particles in the atmosphere. To ensure stable moving object detection, an image restoration process with haze removal must be accompanied. Therefore, this paper proposes an image dehazing network that estimates the current weather conditions and removes haze using the haze level to improve the detection performance under poor weather conditions due to haze and low visibility. Combined with the thermal image, the restored image is assigned to the two You Only Look Once (YOLO) object detectors, respectively, which detect moving targets independently and improve object detection performance using late fusion. The proposed model showed improved dehazing performance compared with the existing image dehazing models and has proved that images taken under foggy conditions, the poorest weather for autonomous driving, can be restored to normal images. Through the fusion of the RGB image restored by the proposed image dehazing network with thermal images, the proposed model improved the detection accuracy by up to 22% or above in a dense haze environment like fog compared with models using existing image dehazing techniques.

AIDM-Strat: Augmented Illegal Dumping Monitoring Strategy through Deep Neural Network-Based Spatial Separation Attention of Garbage.

Economic and social progress in the Republic of Korea resulted in an increased standard of living, which subsequently produced more waste. The Korean government implemented a volume-based trash disposal system that may modify waste disposal characteristics to handle vast volumes of waste efficiently. However, the inconvenience of having to purchase standard garbage bags on one's own led to passive participation by citizens and instances of illegally dumping waste in non-standard plastic bags. As a result, there is a need for the development of automatic detection and reporting of illegal acts of garbage dumping. To achieve this, we suggest a system for tracking unlawful rubbish disposal that is based on deep neural networks. The proposed monitoring approach obtains the articulation points (joints) of a dumper through OpenPose and identifies the type of garbage bag through the object detection model, You Only Look Once (YOLO), to determine the distance of the dumper's wrist to the garbage bag and decide whether it is illegal dumping. Additionally, we introduced a method of tracking the IDs issued to the waste bags using the multi-object tracking (MOT) model to reduce the false detection of illegal dumping. To evaluate the efficacy of the proposed illegal dumping monitoring system, we compared it with the other systems based on behavior recognition. As a result, it was validated that the suggested approach had a higher degree of accuracy and a lower percentage of false alarms, making it useful for a variety of upcoming applications.

Improved Monitoring of Wildlife Invasion through Data Augmentation by Extract-Append of a Segmented Entity.

Owing to the continuous increase in the damage to farms due to wild animals' destruction of crops in South Korea, various methods have been proposed to resolve these issues, such as installing electric fences and using warning lamps or ultrasonic waves. Recently, new methods have been attempted by applying deep learning-based object-detection techniques to a robot. However, for effective training of a deep learning-based object-detection model, overfitting or biased training should be avoided; furthermore, a huge number of datasets are required. In particular, establishing a training dataset for specific wild animals requires considerable time and labor. Therefore, this study proposes an Extract-Append data augmentation method where specific objects are extracted from a limited number of images via semantic segmentation and corresponding objects are appended to numerous arbitrary background images. Thus, the study aimed to improve the model's detection performance by generating a rich dataset on wild animals with various background images, particularly images of water deer and wild boar, which are currently causing the most problematic social issues. The comparison between the object detector trained using the proposed Extract-Append technique and that trained using the existing data augmentation techniques showed that the mean Average Precision (mAP) improved by ≥2.2%. Moreover, further improvement in detection performance of the deep learning-based object-detection model can be expected as the proposed technique can solve the issue of the lack of specific data that are difficult to obtain.

RGDiNet: Efficient Onboard Object Detection with Faster R-CNN for Air-to-Ground Surveillance.

An essential component for the autonomous flight or air-to-ground surveillance of a UAV is an object detection device. It must possess a high detection accuracy and requires real-time data processing to be employed for various tasks such as search and rescue, object tracking and disaster analysis. With the recent advancements in multimodal data-based object detection architectures, autonomous driving technology has significantly improved, and the latest algorithm has achieved an average precision of up to 96%. However, these remarkable advances may be unsuitable for the image processing of UAV aerial data directly onboard for object detection because of the following major problems: (1) Objects in aerial views generally have a smaller size than in an image and they are uneven and sparsely distributed throughout an image; (2) Objects are exposed to various environmental changes, such as occlusion and background interference; and (3) The payload weight of a UAV is limited. Thus, we propose employing a new real-time onboard object detection architecture, an RGB aerial image and a point cloud data (PCD) depth map image network (RGDiNet). A faster region-based convolutional neural network was used as the baseline detection network and an RGD, an integration of the RGB aerial image and the depth map reconstructed by the light detection and ranging PCD, was utilized as an input for computational efficiency. Performance tests and evaluation of the proposed RGDiNet were conducted under various operating conditions using hand-labeled aerial datasets. Consequently, it was shown that the proposed method has a superior performance for the detection of vehicles and pedestrians than conventional vision-based methods.

DBSCAN-Based Tracklet Association Annealer for Advanced Multi-Object Tracking.

Recently, as the demand for technological advancement in the field of autonomous driving and smart video surveillance is gradually increasing, considerable progress in multi-object tracking using deep neural networks has been achieved, and its application field is also expanding. However, various problems have not been fully addressed owing to the inherent limitations in video cameras, such as the tracking of objects in an occluded environment. Therefore, in this study, we propose a density-based object tracking technique redesigned based on DBSCAN, which has high robustness against noise and is excellent for nonlinear clustering. Moreover, it improves the noise vulnerability inherent to multi-object tracking, reduces the difficulty of trajectory separation, and facilitates real-time processing through simple structural expansion. Through performance test evaluation, it was confirmed that by using the proposed technique, several performance indices were improved compared to the existing tracking technique. In particular, when added as a post processor to the existing tracker, the tracking performance owing to noise suppression was considerably improved by more than 10%. Thus, the proposed method can be applied in industrial environments, such as real pedestrian analysis and surveillance security systems.

Delaunay Triangulation-Based Spatial Clustering Technique for Enhanced Adjacent Boundary Detection and Segmentation of LiDAR 3D Point Clouds.

In spatial data with complexity, different clusters can be very contiguous, and the density of each cluster can be arbitrary and uneven. In addition, background noise that does not belong to any clusters in the data, or chain noise that connects multiple clusters may be included. This makes it difficult to separate clusters in contact with adjacent clusters, so a new approach is required to solve the nonlinear shape, irregular density, and touching problems of adjacent clusters that are common in complex spatial data clustering, as well as to improve robustness against various types of noise in spatial clusters. Accordingly, we proposed an efficient graph-based spatial clustering technique that employs Delaunay triangulation and the mechanism of DBSCAN (density-based spatial clustering of applications with noise). In the performance evaluation using simulated synthetic data as well as real 3D point clouds, the proposed method maintained better clustering and separability of neighboring clusters compared to other clustering techniques, and is expected to be of practical use in the field of spatial data mining.

Improvement of Anomalous Behavior Detection of GNSS Signal Based on TDNN for Augmentation Systems.

The reliability of a navigation system is crucial for navigation purposes, especially in areas where stringent performance is required, such as civil aviation or intelligent transportation systems (ITSs). Therefore, integrity monitoring is an inseparable part of safety-critical navigation applications. The receiver autonomous integrity monitor (RAIM) has been used with the global navigation satellite system (GNSS) to provide integrity monitoring within avionics itself, such as in civil aviation for lateral navigation (LNAV) or the non-precision approach (NPA). However, standard RAIM may not meet the stricter aviation availability and integrity requirements for certain operations, e.g., precision approach flight phases, and also is not sufficient for on-ground vehicle integrity monitoring of several specific ITS applications. One possible way to more clearly distinguish anomalies in observed GNSS signals is to take advantage of time-delayed neural networks (TDNNs) to estimate useful information about the faulty characteristics, rather than simply using RAIM alone. Based on the performance evaluation, it was determined that this method can reliably detect flaws in navigation satellites significantly faster than RAIM alone, and it was confirmed that TDNN-based integrity monitoring using RAIM is an encouraging alternative to improve the integrity assurance level of RAIM in terms of GNSS anomaly detection.

Development of an RTK-GPS positioning application with an improved position error model for smartphones.

This study developed a smartphone application that provides wireless communication, NRTIP client, and RTK processing features, and which can simplify the Network RTK-GPS system while reducing the required cost. A determination method for an error model in Network RTK measurements was proposed, considering both random and autocorrelation errors, to accurately calculate the coordinates measured by the application using state estimation filters. The performance evaluation of the developed application showed that it could perform high-precision real-time positioning, within several centimeters of error range at a frequency of 20 Hz. A Kalman Filter was applied to the coordinates measured from the application, to evaluate the appropriateness of the determination method for an error model, as proposed in this study. The results were more accurate, compared with those of the existing error model, which only considered the random error.

Self-organizing maps with dynamic learning for signal reconstruction.

Wireless Brain Machine Interface (BMI) communication protocols are faced with the challenge of transmitting the activity of hundreds of neurons which requires large bandwidth. Previously a data compression scheme for neural activity was introduced based on Self Organizing Maps (SOM). In this paper we propose a dynamic learning rule for improved training of the SOM on signals with sparse events which allows for more representative prototype vectors to be found, and consequently better signal reconstruction. This work was developed with BMI applications in mind and therefore our examples are geared towards this type of signals. The simulation results show that the proposed strategy outperforms conventional vector quantization methods for spike reconstruction.

Modeling and inverse controller design for an unmanned aerial vehicle based on the self-organizing map.

The next generation of aircraft will have dynamics that vary considerably over the operating regime. A single controller will have difficulty to meet the design specifications. In this paper, a self-organizing map (SOM)-based local linear modeling scheme of an unmanned aerial vehicle (UAV) is developed to design a set of inverse controllers. The SOM selects the operating regime depending only on the embedded output space information and avoids normalization of the input data. Each local linear model is associated with a linear controller, which is easy to design. Switching of the controllers is done synchronously with the active local linear model that tracks the different operating conditions. The proposed multiple modeling and control strategy has been successfully tested in a simulator that models the LoFLYTE UAV.

Effects of Different Salts on Salt-Assisted Ultrasonic Spray Pyrolysis (SA-USP) Calcination for the Synthesis of Strontium Ferrite.

Strontium ferrite (hexaferrite), SrFe12O19, was successfully fabricated in sizes ranging from hundreds of nanometers to several micrometers by salt-assisted ultrasonic spray pyrolysis-calcination using different salt media. All samples were single phases of SrFe12O19 without the intermediate phase, α-Fe2O3, and their morphology was hexagonal. As calcination temperature increased, the size of as-calcined samples and saturation magnetization, Ms, increased while coercivity decreased. The particle size of the obtained nanoparticles varied depending on the salt media and calcination temperatures. The best magnetic properties obtained in this experiment were a coercivity of 6973 Oe with a saturation magnetization of 68.3 emu/g. To the best of our knowledge, these coercivity values are the highest ever obtained. We propose a detailed mechanism explaining the growth of these particles and conclude that the resulting single-domain particle size is about 70 nm, taking into account of factors affecting coercivity in ferrite nano- to micro-sized particles.