Yazar "Kayaarma, Selma Yilmazyildiz" seçeneğine göre listele
Listeleniyor 1 - 5 / 5
Sayfa Başına Sonuç
Sıralama seçenekleri
Öğe A Near-Real Time Automatic Audio Classification: Special Case for Hacivat and Karagoz Shadow Play(Institute of Electrical and Electronics Engineers Inc., 2025) Sevdi, Onur Eren; Topaloglu, Yakup; Kayaarma, Selma YilmazyildizSpeaker identification plays a key role in various applications, such as security, biometrics, and human-computer interaction. As a specific task under the domain of audio classification, speaker identification aims to recognize individuals based on their voice characteristics. This paper presents a comparison between three widely adopted neural network architectures and evaluates their performance as classifiers for a real-time speaker identification system. A custom-collected dataset was gathered using publicly shared YouTube videos of a single speaker imitating multiple characters from traditional Turkish shadow play Karagoz and Hacivat. Both MFCC and Log-Mel filterbank energy features were used during the training of CRNN, 2D-CNN and Bi-LSTM architectures. Among these architectures, 2D-CNN achieved the highest accuracy with a value of 94.4% and was approximately 2.7 times faster than its closest follower Bi-LSTM during real-time testing on RTX 4070 Super GPU. © 2025 IEEE.Öğe Deep Learning Based Growth Analysis and Disease Detection in Strawberry Cultivation(Institute of Electrical and Electronics Engineers Inc., 2023) Ayberguler, Azad; Arslan, Enis; Kayaarma, Selma YilmazyildizStrawberry cultivation can be susceptible to unforeseen diseases. For the prevention of Powdery Mildew and Gray Mold diseases prompt pesticide applications should be carried out during the disease development or periodically, aligned with strawberries' growth stages. In this study, a deep learning-based growth stage analysis and disease detection solution was developed. Three versions of the YOLO architecture (YOLOv5, YOLOv3, YOLOv3-tiny) have been trained on a dataset that was enhanced for this specific use case. YOLOv3-tiny version was also deployed on a simple unmanned ground vehicle in a real-life strawberry cultivation greenhouse. This study differs from others in the literature by training and deploying a model that would enable the detection of both powdery mildew disease and the growth stages of strawberries in a single model. With the deployment of this model, the strawberry growers can implement the appropriate spraying strategies that would control and prevent the formation and spread of these diseases. © 2023 IEEE.Öğe Deep Learning-Based Classification Of Harmandali Dance Figures(Institute of Electrical and Electronics Engineers Inc., 2025) Buz, Burakhan; Kayaarma, Selma YilmazyildizThis study aims to recognize figures from the Turkish folk dance Harmandali using deep learning-based methods. Although AI-based studies on our country's folk dances are quite rare, this study is one of the first examples integrating robotic systems. Accordingly, a special video dataset consisting of Harmandali figures was created; the images extracted from the videos were processed using the Google MediaPipe Pose library, and skeleton keypoints representing the dancer's joint positions were extracted from each frame. The resulting time series data were classified using 1D-CNN, LSTM, and GRU architectures, and their performances were compared. Experimental results show that GRU-based models achieve the highest success with 89.7% top-1, 97.58% top-3 accuracy rates and 0.8953 F1 score. This study demonstrates that deep learning approaches based on skeleton representations are effective for the automatic recognition of dance figures. © 2025 IEEE.Öğe Facial Emotion Recognition for Imitation in Human-Robot Interaction(Institute of Electrical and Electronics Engineers Inc., 2024) Yanç, Ibrahim; Ipek, Aykan; Kayaarma, Selma YilmazyildizThe importance of human-robot interaction (HRI) is increasing day by day and new studies are being carried out every day. In this study, an animatronic robot face that detects human emotions and can mimic these emotions was designed to increase human-robot interaction. Thanks to rapidly developing technological competencies, serious progress has been made in artificial intelligence and scientific studies in this field. Human emotion analysis plays a vital role in human-robot interaction. Since the ability of traditional algorithms to analyze and classify emotions from real-time images is below the acceptable threshold, machine learning and artificial neural networks methods have emerged. In this work, an original animatronic robot face with 18 degrees of freedom was designed. By training a custom Convolutional Neural Network (CNN) model with a mixed dataset consisting of Fer2013, CKPlus, and Kdef datasets, an accuracy rate of 72% was achieved. This trained model was integrated into the system to detect the emotion from the human face and to mimic the detected emotion. © 2024 IEEE.Öğe Passenger Density Detection in Railway Carriages(Institute of Electrical and Electronics Engineers Inc., 2024) Aras, Yusuf Efe; Akpınar, Muhammet; Kayaarma, Selma YilmazyildizIn this study, the goal is to contribute to reducing waiting times and to improve passanger comfort with passenger density estimation using deep learning methods. For this purpose, the YOLOv8 deep learning model was used to detect passenger density in urban rail systems. The model is trained with the CrowdHuman dataset. The trained model runs on a Raspberry Pi 5 and processes images obtained from IP cameras. These processed images are stored in an SQL Server via an API and the density estimation results are displayed on an LCD screen. This design aims to make the system feasible in the field in terms of performance and cost. The trained model, detects the passenger occupancy with a test accuracy rate of up to 90% and offers significant advantages in real-time applications due to its low computational power requirements. © 2024 IEEE.
