FOREIGN: ADAFACE: QUALITY ADAPTIVE MARGIN FOR FACE RECOGNITION Authors: Minchul Kim, Anil K. Jain, Xiaoming Liu Department of Computer Science and Engineering, Michigan State University, East Lansing, MI, 48824 Published in CVPR2022 (Oral) https://openaccess.thecvf.com/content/CVPR2022/html/Kim_AdaFace_Quality_Adaptive_Margi n_for_Face_Recognition_CVPR_2022_paper.html Recognition in low quality face datasets is challenging because facial attributes are obscured and degraded. Advances in margin-based loss functions have resulted in enhanced discriminability of faces in the embedding space. Further, previous studies have studied the effect of adaptive losses to assign more importance to misclassified (hard) examples. In this work, we introduce another aspect of adaptiveness in the loss function, namely the image quality. We argue that the strategy to emphasize misclassified samples should be adjusted according to their image quality. Specifically, the relative importance of easy or hard samples should be based on the sample's image quality. We propose a new loss function that emphasizes samples of The study explores AdaFace's technical details, going over how it was implemented and how well it performed in comparison to other face recognition techniques. The study probably sheds light on the difficulties in face recognition and how AdaFace overcomes them to raise the standard of face recognition software as a whole. The study that is being presented presents AdaFace, a novel method for tackling the problem of face identification in datasets with poor quality. The suggested approach makes use of an adaptive loss function that modifies the emphasis on simple or complex instances based on the image quality of face samples. On several datasets, the work shows enhanced performance above the stateof-the-art techniques. These include investigating more reliable measures of image quality, determining whether or not they are generalizable across various datasets, analyzing the effects of image preprocessing, assessing computational effectiveness, improving the interpretability of the adaptive margin function, researching resilience to adversarial attacks, and verifying different difficulties based on their image quality. Our method achieves this in the form of an adaptive margin function by approximating the image quality with feature norms. Extensive experiments show that our method, AdaFace, improves the face recognition performance over the state-of-the-art (SoTA) on four datasets (IJB-B, IJB-C, IJB-S and TinyFace). Code and models are released in Supp. effectiveness in practical applications. Closely examining these gaps may lead to a more thorough comprehension and improvement of adaptive loss functions for face recognition under difficult circumstances. FOREIGN: Face Recognition: A Literature Review Barnouti, N. H. (n.d.). Face Recognition: A Literature Review. https://www.ijais.org/archives/volume11/number4/935-2016451597/ Authors: Nawaf Hazim Barnouti, Sinan Sameer Mahmood Al-dabbagh, Wael Esam Matti Year of Publication: 2016 https://www.ijais.org/archives/volume11/number4/935-2016451597/ Face recognition have gained a great deal of popularity because of the wide range of applications such as in entertainment, smart cards, information security, law enforcement, and surveillance. It is a relevant subject in pattern recognition, computer vision, and image processing. Two major methods are used for features extraction, which can be classified into appearancebased and Model-based methods. Appearance-based methods use global representations to identify a face. Model-based face methods aim to construct a model of the human face that capture facial variations. Image similarity is the distance between the vectors of two images. This paper contains Four sections. The first section discusses face recognition applications with examples. The second section discuss the common feature face recognition methods. The third section discuss distance measurement classifiers. The fourth section The researchers highlights the use of facial recognition in smart cards, entertainment, information security, law enforcement, and surveillance while talking about the technology's uses. Appearance-based and model-based feature extraction techniques are the two main approaches identified. While modelbased approaches seek to build a model incorporating facial variability, appearancebased approaches use global representations for face identification. Although facial recognition techniques are introduced, there is a lack of comprehensive performance evaluation and adaption assessments for practical situations in the literature. It is unknown how resilient model-based techniques are to changes in facial features. There is no talk about potential biases in face recognition databases, and emerging technologies and ethical issues are disregarded. Filling such these gaps is essential to gaining a more thorough and useful understanding of face recognition technologies. discuss different face recognition databases. FOREIGN: Deep-Learning-Enhanced Multitarget Detection for End–Edge– Cloud Surveillance in Smart IoT Authors: X. Zhou, X. Xu, W. Liang, Z. Zeng and Z. Yan, "Deep-Learning-Enhanced Multitarget Detection for End–Edge–Cloud Surveillance in Smart IoT," in IEEE Internet of Things Journal, vol. 8, no. 16, pp. 12588-12596, 15 Aug.15, 2021, doi: 10.1109/JIOT.2021.3077449 Published in: IEEE Internet of Things Journal ( Volume: 8, Issue: 16, 15 August 2021) https://ieeexplore.ieee.org/abstract/document/9422817/authors#authors Along with the rapid development of cloud computing, IoT, and AI technologies, cloud video surveillance (CVS) has become a hotly discussed topic, especially when facing the requirement of real-time analysis in smart applications. Object detection usually plays an important role for environment monitoring and activity tracking in surveillance system. The emerging edge-cloud computing paradigm provides us an opportunity to deal with the continuously generated huge amount of surveillance data in an on-site manner across IoT systems. However, the detection performance is still far away from satisfactions due to the complex surveilling environment. In this study, we focus on the multitarget detection for real-time surveillance in smart IoT systems. A newly designed deep neural network model Using the A-YONet model, a hybrid of YOLO and MTCNN, this study improves multitarget identification in smart IoT systems, which is important for real-time monitoring in smart applications. By utilizing edge-cloud computing for minimal instruction, the suggested approach enhances accuracy in difficult surveillance settings. Its usefulness is confirmed by experiments on real and public datasets, highlighting its significance in improving real-time surveillance for intelligent Internet of Things applications. There are several limitations in our knowledge of AYONet's practical usability and efficacy because the study on its multitarget detection in smart IoT systems does not investigate its scalability, adaptability, robustness, ethical implications, or comparative performance comparison with existing models. For practical implementation in a variety of monitoring contexts, closing these gaps is essential. called A-YONet, which is constructed by combining the advantages of YOLO and MTCNN, is proposed to be deployed in an end-edgecloud surveillance system, in order to realize the lightweight training and feature learning with limited computing sources. An intelligent detection algorithm is then developed based on a preadjusting scheme of anchor box and a multilevel feature fusion mechanism. Experiments and evaluations using two data sets, including one public data set and one homemade data set obtained in a real surveillance system, demonstrate the effectiveness of our proposed method in enhancing training efficiency and detection precision, especially for multitarget detection in smart IoT application developments. Vision-based Crowd Counting and Social Distancing Monitoring using Tiny-YOLOv4 and DeepSORT Author’s J. C. Valencia, E. P. Dadios, A. M. Fillone, J. C. V. Puno, R. G. Baldovino and R. K. C. Billones, "Vision-based Crowd Counting and Social Distancing Monitoring using Tiny-YOLOv4 and DeepSORT," 2021 IEEE International Smart Cities Conference (ISC2), Manchester, United Kingdom, 2021, pp. 1-7, doi: 10.1109/ISC253183.2021.9562868. https://ieeexplore.ieee.org/abstract/document/9562868 With the novel coronavirus, social distancing and crowd monitoring became vital in managing the spread of the virus. This paper presents a desktop application that utilizes Tiny-YOLOv4 and DeepSORT tracking algorithm to monitor crowd count and social distancing in a top-view camera perspective. The application is able to process video files or live camera feed such as CCTV or surveillance cameras and generate reports indicating people detected per unit time, percentage of social distancing per unit time, detection and social distancing logs as well as color-coded bounding boxes to indicate if the detected people are following social distancing protocols. This work, which presents a desktop program for crowd surveillance and social distance using Tiny-YOLOv4 and the DeepSORT tracking algorithm, is extremely pertinent in light of the COVID-19 pandemic. After processing video files or live camera feeds, the application generates detailed logs, social distance percentages, and reports on people it has detected. Color-coded bounding boxes are a useful tool for crowd control in realtime during public health emergencies because they visually signal adherence to social distancing standards. There are some insufficient studies in the article on a desktop program for social distancing and COVID-19 crowd surveillance. There are no discussions on the accuracy and dependability of the Tiny-YOLOv4 and DeepSORT tracking algorithms, adaptation of the algorithm to various surveillance contexts, or performance in real-world scenarios. It is lacking the implementation of privacypreserving measures and addressing privacy concerns. Furthermore, a vacuum in knowledge regarding the performance of the suggested application remains due to the lack of a comparative analysis with current solutions. Filling in these gaps is essential to a thorough assessment of the applicability's accuracy, practicality, and ethical implications. LOCAL: The Technology Adoption and Governance of Artificial Intelligence in the Philippines Author’s Ronnie S. Concepcion Electronics and Communications Engineering Department, De La Salle University, Manila, Philippines Rhen Anjerome R. Bedruz Manufacturing Engineering and Management Department, De La Salle University, Manila, Philippines Alvin B. Culaba Mechanical Engineering Department, De La Salle University, Manila, Philippines Elmer P. Dadios Manufacturing Engineering and Management Department, De La Salle University, Manila, Philippines Athena Rosz Ann R. Pascua Electronics and Communications Engineering Department, De La Salle University, Manila, Philippines https://ieeexplore.ieee.org/abstract/document/9072725 Artificial intelligence is primed to disrupt our society and the industry. The AI trend of technological singularity is continuously accelerating and is being employed to the different facets of humanity from education, medicine, business, engineering, arts and the like. Government and private companies have been hooked up with this fast pacing technology. AI may displace some non-digital jobs that performs heavy load and repetitive tasks, but it certainly augments labor shortage by realigning the workforce competitiveness to what the technology requires. Artificial Intelligence is rapidly changing education, healthcare, business, and other sectors of society and economy. It addresses labor shortages and increases worker competitiveness, even though it might eliminate some employment. The Philippines can take the lead in the world in AI adoption by putting the PDP 2017– 2022 and its HNRDA policies into practice. Although there are a number of research gaps, the summary highlights how AI has the potential to revolutionize the Philippines. It doesn't go into detail on possible job displacement, doesn't look into adoption barriers, and doesn't provide particular instances of how AI affects productivity. Furthermore, it highlights the significance of technological governance without getting into specific difficulties. Additional investigation may yield valuable perspectives on the societal implications, obstacles to adoption, productivity measures, governance issues, and The diffusion of AI technology is necessary for mental shift of the government and industry leaders to adopt the technology. Research and development is very promising to uplift mankind to faster productivity and positively affect the industries in international perspective. The Philippines is still coping up with the adoption of AI system, but it can steer up globally by strengthening the technology governance of strictly implementing the policies with measures the PDP 2017-2022 and its HNRDA. factual data linking the use of AI to worldwide leadership. For a comprehensive grasp of AI's consequences in the Philippines and its place in the world at large, these gaps must be filled. Design of face detection and recognition system for smart home security application Author’s: Dwi Ana Ratna Wati Universitas Islam Indonesia, Yogyakarta, Daerah Istimewa Yogyakart, ID Dika Abadianto Department of Electrical Engineering, Universitas Islam Indonesia, Yogyakarta, Indonesia Published in: 2017 2nd International conferences on Information Technology, Information Systems and Electrical Engineering (ICITISEE) https://ieeexplore.ieee.org/abstract/document/8285524 This research designs face detection and recognition systems for smart home security application. The design is implemented using MyRIO 1900 and programmed using LabVIEW. The connection between myRIO and computer is wifi network. The image of a person is acquired via webcam connected to MyRIO using USB cable. The face detection system is built based on the template matching, while the face recognition is based on the principle component analysis. The testing is done to examine the performance of the face detection in various change of distance, light intensity, light position angles, person's accessories and shirt colour. The face detection modul has good performance in some conditions as distance between the person and the camera is less than 240 cm, person doesn't use accessories that cover part of face, person doesn't use shirt with colour similar to skin colour, and background colour is difference from skin colour. While the face recognition system has 80% of accuracy when it is tested using realtime image. The combination with password is needed in order to increase Because it creates face detection and recognition algorithms specifically for smart home security applications, this research is extremely pertinent. The design connects via wifi and uses LabVIEW and MyRIO 1900 to capture photographs of people using a webcam. Principal component analysis is the foundation for face recognition, whereas template matching is used by the face detection system. Testing demonstrates good performance within predetermined boundaries under a variety of situations. In tests conducted in real time, the facial recognition system reaches an accuracy of 80%. Since adding a password improves security, this research is useful and applicable to actual smart home security systems. Although the system performs well in certain scenarios, there are no specifics about possible testing constraints in the abstract. Further investigation is required into the system's scalability and adaptability to various hardware and software settings. The robustness of the algorithm to changes in facial expressions, positions, and environmental conditions is not covered in the abstract, and there are no details provided regarding the dataset that was used to assess the accuracy of face recognition. Moreover, further research is necessary to identify any potential weaknesses and concerns with the suggested passwordprotected facial recognition system. In real-world smart home security systems, closing these gaps is crucial to improving the system's security, scalability, and dependability. the security level as it is applied in real smart home security systems.