眼鏡の着脱にロバストな高速顔画像同定方式

India’s defense drone ecosystem remains largely dependent on manual control and pre-programmed routines, restricting adaptability in rapidly evolving or GPS-denied environments. In this paper, we propose Drishti-AI Drone-based Reinforcement System for Intelligent Human Tracking and Identification using AI , a fully autonomous UAV surveillance framework that tightly integrates Reinforcement Learning (RL)–based navigation with real-time Deep Learning (DL)–driven human detection. For autonomous mobility, we introduce AeroNav-PPO, a Proximal Policy Optimization (PPO) model trained in a 3D AirSim environment to enable intelligent navigation and dynamic obstacle avoidance. For perception, we present YOLOv8-HawkEye, a customized object detection model capable of identifying humans with real-time coordinate mapping and automated alert generation. Together, these modules form a closed perception-action loop that enables end-to-end autonomous operation. The AeroNav-PPO agent demonstrated an average episode reward of 347.0 with consistent patrol performance across complex terrains, while YOLOv8-HawkEye achieved a precision of 95.6%, recall of 87.8%, and mAP@0.5 of 93.8%. Our results validate Drishti-AI as a promising step toward fully autonomous, mission-ready surveillance drones for next-generation defense applications.

In this paper, we present an automatic face replacement approach in photographs based on Active Shape Models (ASM). Our replacement algorithm has three main modules: face alignment, face morph, and seamless blending. First, given an input and target image, we automatically implement face alignment using ASM. Second, we fit a morphable 2D model to the input face, and warp its shape to match the target face shape. Last, we adjust the lighting and color of the target face image to match those of the input face image, and seamlessly blend the target face image into the input image. Our approach is 2D model-based, fully automatic, and generates natural results. The improved results demonstrate the effectiveness of our approach.

Lack of identity documents is an important obstacle to the protection of people forced to leave their homes by conflict, persecution, or natural disaster. At the most basic level, a person lacking identity documents cannot travel through the legal channels. Lack of identification can make people more vulnerable to trafficking, for example by making it more difficult to prove a person’s age or family relationships. Those who lack identity documents may face greater difficulties proving their entitlement to nationality or to refugee status. The absence of fair and effective processes for registration and identification places displaced persons at the risk of exploitation and exclusion. Access to essential public services will be limited. Children are at risk of becoming stateless because of difficulties in accessing birth registration and because the rules and practices in place may prevent them from acquiring the nationality either of (one of) their parents or of the country of birth. Poorly implemented identification systems may even put displaced people at greater risk. If registration and identification procedures are not properly carried out, people who qualify to be recognized as refugees or stateless. Persons, or as nationals, may be subject to immigration detention and deportation to a country where they are in danger, or to illegal exploitation where they are. Weak identification systems can make it difficult for displaced persons and their children to reunite, to repatriate after crises have ended, and to reclaim land and property that they left behind. Robust identification and registration systems for displaced people also mitigate the disruptive impacts of rapid influxes of refugees for governments, and assist planning to respond to the needs of the displaced populations.

The classical language identification architecture would require a collection of languages independent text and speech information for training by the system before it can identify the languages correctly. This paper also address language identification framework however with data has been downsized considerably from general language identification architecture. The system goal is to identify the type of language being spoken to the system based on a series of trained speech with sound file features and without any language text data or lexical knowledge of the spoken language. The system is also expected to be able to be deployed in mobile platform in future. This paper is specifically about measuring the performance optimisation of audio filters on a CNN model integration for the language identification system. There are several metric to gauge the performance identification system for a classification problem. Precision, recall and F1 Scores is presented for the performance evaluation with different combination of filters together with CNN model as the framework of the language identification system. The goal is not to get the best filter for noise, instead to identify the filter that is a good fit to develop language model with environmental noise for a robust language identification system. Our experiments manage to identify the best combination of filters to increase the accuracy of language identification using short speech. This resulting us to modify our pre-processing phase in the overall language identification system.

This paper presents a wireless distributed pyroelectric sensor system for tracking and identifying multiple humans based on their body heat radiation. This study aims to make pyroelectric sensors a low-cost alternative to infrared video sensors in thermal gait biometric applications. In this system, the sensor field of view (FOV) is specifically modulated with Fresnel lens arrays for functionality of tracking or identification, and the sensor deployment is chosen to facilitate the process of data-object-association. An Expectation-Maximization-Bayesian tracking scheme is proposed and implemented among slave, master, and host modules of a prototype system. Information fusion schemes are developed to improve the system identification performance for both individuals and multiple subjects. The fusion of thermal gait biometric information measured by multiple nodes is tested at four levels: sample, feature, score, and decision. Experimentally, the prototype system is able to simultaneously track two individuals in both follow-up and crossover scenarios with average tracking errors less than 0.5 m. The experimental results also demonstrate system's potential to be a reliable biometric system for the verification/identification of a small group of human subjects. The developed wireless distributed infrared sensor system can run as a standalone prisoner/patient monitoring system under any illumination conditions, as well as a complement for conventional video and audio human tracking and identification systems.

ECG (Electrocardiogram) data analysis is one of the most widely used and important tools in cardiology diagnostics. In recent years the development of advanced deep learning techniques and GPU hardware have made it possible to train neural network models that attain exceptionally high levels of accuracy in complex tasks such as heart disease diagnoses and treatments. We investigate the use of ECGs as biometrics in human identification systems by implementing state-of-the-art deep learning models. We train convolutional neural network models on approximately 81k patients from the US, Germany and China. Currently, this is the largest research project on ECG identification. Our models achieved an overall accuracy of 95.69%. Furthermore, we assessed the accuracy of our ECG identification model for distinct groups of patients with particular heart conditions and combinations of such conditions. For example, we observed that the identification accuracy was the highest (99.7%) for patients with both ST changes and supraventricular tachycardia. We also found that the identification rate was the lowest for patients diagnosed with both atrial fibrillation and complete right bundle branch block (49%). We discuss the implications of these findings regarding the reidentification risks of patients based on ECG data and how seemingly anonymized ECG datasets can cause privacy concerns for the patients.

Development of biometric human identification algorithms and systems is reviewed. A crisis that prevents them from being used in large computing systems is demonstrated. Comprehensive reliable tests are considered as a way to overcome the crisis. Ways to take into account the influence various factors may have on the reliability of evaluating the practical efficiency of remote human identification algorithms and systems are addressed. A method to determine and calculate quantitative quality criteria of multimodal biometric data is proposed; the possibility of extrapolating the results of testing biometric algorithms to practical application is discussed. Quality functions and the artificiality of biometric data given as the measure of proximity to biometric data registered from an unaware recognition subject are studied.

The demand for safety-boosting systems is increasing, especially nowadays, to limit the rapid spread of COVID-19. Real-time life-logging is an essential application towards tracking infected cases and thus containing the pandemic outbreak. This application raises the need for an accurate human identification technology where cameras cannot be adopted due to privacy. Recently, LiDAR sensor has attracted attention due to its ability to represent the surrounding environment in the form of 3D point cloud. In this paper, we introduce a novel wearable device of a micro-size LiDAR to build an onboard human identification system for life-logging. The system acquires 3D point cloud data of the surrounding environment from which subject-discriminative signatures are extracted. This is achieved by removing noise and background using Spatio-temporal density clustering and fisher vector representations. The extracted features are then used to train a random forest classifier for subject identification. We have implemented and evaluated the proposed system on six different subjects. The results show that the proposed system can effectively remove noise and background and accurately identify subjects with 99.9% accuracy.

Identity recognition in real environment with reliable mode is a key issue in human computer interaction (HCI). In this paper, we present a robust speaker identification system considering score based optimal reliability measure of different modalities. We propose an extension of the modified convection function's optimizing parameter to account optimal reliability simultaneously via audio and lip information based reliability measure in bimodal speaker identification system for robust speaker identification. For degradation of visual signals, we have applied JPEG compression to test images. In addition, for creating mismatch in between enrollment and test session, acoustic Babble noises and artificial illumination have been added to test audio and visual signals, respectively. Local PCA has been used to both modalities for reducing the dimension of feature vector. We have applied a swarm intelligence algorithm i.e., particle swarm optimization for optimizing the modified convection function's optimizing parameters. The overall speaker identification experiments are performed using VidTimit DB. Experimental results show that our proposed optimal reliability measures have effectively enhanced the identification accuracy of 7.73% in comparison with the best classifier system in the integration system and maintains the modality reliability statistics in term of its performance thus verified the consistency of the proposed extension.

Brain-computer interfaces (BCIs) offer a direct pathway between neural activity and computational systems, enabling enhanced human performance. We developed a BCI system for real-time human target identification in aerial imagery (5 Hz), using Event-Related Potentials (ERPs) related to visual attention processes, specifically the N2pc and P3. Six subjects participated in two distinct Rapid Serial Visual Presentation (RSVP) paradigms: one for training/validation and the other for testing with novel environments and target appearances. Three ERP-based classification models were evaluated: N2pc, P3, and a combined N2pc and P3 approach. The combined model achieved the highest target identification in validation (AUC = 0.87) but declined on novel stimuli (AUC = 0.75). N2pc-only models maintained consistent performance across both sets (AUC = 0.77 validation, 0.75 testing), demonstrating strong generalization. Optimizing model selection and channel configuration per subject yielded an average sensitivity of 0.79 and specificity of 0.76 for target vs. non-target classification, respectively. For lateralized target detection, N2pc-based models classified Left vs. Right appearances with an AUC of 0.75 in both datasets. Target centroid prediction achieved correlation scores of 0.49 (validation) and 0.52 (testing) with the ground truth, indicating reliable spatial localization. These results validate the N2pc as a robust, generalizable neural marker for visual target detection and localization, enabling gaze-independent, multi-target discrimination, making it suitable for real-world search and rescue operations. This work provides a proof of concept for ERP-based BCI systems that maintain high performance across varied environments.

Biometric authentication for personal identification is very popular now a days. Human ear recognition system is a new technology in this field. The change of appearance with the expression was a major problem in face biometrics but in case of ear biometrics the shape and appearance is fixed. That is why it is advantageous to use it for personal identification. In this paper, we have proposed a new approach for an automated system for human ear identification. Our proposed method consists of three stages. In the first stage, preprocessing of ear image is done for its contrast enhancement and size normalization. In the second stage, features are extracted through Haar wavelets followed by ear identification using fast normalized cross correlation in the third stage. The proposed method is applied on USTB ear image database and IIT Delhi. Experimental results show that our proposed system achieves an average accuracy of 97.2% and 95.2% on these databases respectively.

In this paper, we show the existence of human radio biometrics and present a human identification system that can discriminate individuals even through the walls in a non-line-of-sight condition. Using commodity Wi-Fi devices, the proposed system captures the channel state information (CSI) and extracts human radio biometric information from Wi-Fi signals using the time-reversal (TR) technique. By leveraging the fact that broadband wireless CSI has a significant number of multipaths, which can be altered by human body interferences, the proposed system can recognize individuals in the TR domain without line-of-sight radio. We built a prototype of the TR human identification system using standard Wi-Fi chipsets with $3 \times 3$ multi-in multi-out (MIMO) transmission. The performance of the proposed system is evaluated and validated through multiple experiments. In general, the TR human identification system achieves an accuracy of 98.78% for identifying about a dozen of individuals using a single transmitter and receiver pair. Thanks to the ubiquitousness of Wi-Fi, the proposed system shows the promise for future low-cost low-complexity reliable human identification applications based on radio biometrics.

This paper presents a further step of a research toward the development of a quick and accurate weapons identification methodology and system. A basic stage of this methodology is the automatic acquisition and updating of weapons ontology as a source of deriving high level weapons information. The present paper outlines the main ideas used to approach the goal. In the next stage, a clustering approach is suggested on the base of hierarchy of concepts. An inherent slot of every node of the proposed ontology is a low level features vector (LLFV), which facilitates the search through the ontology. Part of the LLFV is the information about the object's parts. To partition an object a new approach is presented capable of defining the objects concavities used to mark the end points of weapon parts, considered as convexities. Further an existing matching approach is optimized to determine whether an ontological object matches the objects from an input image. Objects from derived ontological clusters will be considered for the matching process. Image resizing is studied and applied to decrease the runtime of the matching approach and investigate its rotational and scaling invariance. Set of experiments are preformed to validate the theoretical concepts.

A comprehensive human trafficking identification and prediction system using machine learning techniques

Unobtrusive and robust human identification using COTS RFID