Communication is needed to interact and socialize in order to connect with the environment and other people. In general, communication uses spoken or written words. However, in some cases in the community there are some people who cannot communicate verbally due to physical limitations such as deaf and speech impaired. Usually, they use nonverbal communication such as body movements and this communication is commonly referred to as sign language. The sign language method is used to spell or pronounce words. However, not everyone can understand the sign language used by the deaf and mute, so a system or tool is needed to bridge communication between the deaf or mute and normal people. One solution that can be offered is the use of computer technology as a tool to identify sign language. The technology is in the form of an automatic language translator system design with processing input images in the form of letter classes A to E, I, You and I Love You using the Convolutional Neural Network (CNN) architecture which by using this method the accuracy value can reach 99.82%.

This paper presents an optimized Automatic License Plate Recognition (ALPR) system designed for resource-constrained devices, leveraging YOLOv8 for real-time object detection and Optical Character Recognition (OCR) to extract license plate information under challenging conditions such as low-light, motion blur, and occlusions. Unlike traditional ALPR systems that rely on high computational resources, our approach balances detection accuracy, processing speed, and efficiency. The system is evaluated on three benchmark datasets: the Chinese City Parking Dataset (CCPD) with 250,000 images under diverse conditions, the UFPR-ALPR Dataset (Universidade Federal do Paraná, Brazil) containing 4,500 real-world traffic images, and the RodoSol-ALPR Dataset with 20,000 highway surveillance images for high-speed license plate recognition. Among various YOLOv8 variants tested, YOLOv8-s achieved the best performance, with a mean Average Precision (mAP) of 99.3% while sustaining over 30 Frames Per Second (FPS), making it suitable for real-time ALPR applications. Furthermore, image sharpening and contour segmentation techniques improved OCR recognition accuracy by 5.1% under low-light conditions, improving robustness. Comparative analysis against state-of-the-art OCR-based ALPR methods (EasyOCR, FastOCR, and CR-NET) demonstrated that our approach surpasses existing models in both recognition rate and computational efficiency. These findings establish YOLOv8 as a highly effective and deployable solution for intelligent transportation, surveillance, and law enforcement applications requiring real-time license plate recognition with minimal computational overhead.

The growth of internet users in Indonesia experiences a very high increase every year, with the increase in internet users in Indonesia also resulting in many people using social media. Sestyc is a social media application created by a group of millennial children in Indonesia. This study was conducted to analyze the sentiment of users of Sestyc using text data in the form of a review obtained from the Google Play Store. The purpose of this research is to analyze sentiment towards the sestyc application and find the best algorithm for classifying sentiment. The algorithm used in analyzing sentiment in this study consists of Support Vector Machine, Logistic regression, and Naive Bayes. The results of sentiment class labeling on the sestyc review data obtained 8000 reviews with a total of 4719 positive reviews and 3281 negative reviews. The results of this study indicate that the Support Vector Machine algorithm has the highest accuracy value compared to other algorithms, where the Support Vector Machine gets an accuracy value. by 87.81%.

Pandemi COVID-19 terjadi terjadi diseluruh dunia, seluruh aktifitas dan kegiatan menjadi terganggu termasuk aktifitas pendidikan. Seluruh kegiatan pendidikan yang semula dilaksanakan secara offline, harus dilaksanakan secara online. Salah satu cara untuk mempermudah kegiatan belajar mengajar, adalah dengan mengunggah video yang berisi materi-materi pembelajaran ke media sosial seperti Youtube ataupun sejenisnya. Hal ini tentunya membawa perubahan yang sangat signifikan terhadap kegiatan belajar mengajar termasuk kegiatan di SD Muhammadiyah Rabbani Kabupaten Klaten Jawa Tengah. Open Broadcast Software atau OBS merupakan salah satu tools open source yang bisa dimanfaatkan untuk membuat video pembelajaran dengan mudah dan tidak terkoneksi internet. Tujuan dari pelatihan ini adalah tentang tatacara pembuatan OBS. Evaluasi menggunakan kuesioner online dan diolah menggunakan SPSS versi 25. Hasil evaluasi menunjukkan bahwa 60% peserta menjawab sangat setuju bahwa pelatihan mudah dimengerti, pemateri sangat handal dan OBS mudah di implementasikan, sedangkan 30% lainnya menjawab setuju.

Nowadays, social media is a normal thing or even mandatory. Because of that, the evolution of social media has become very fast and has made people not know the boundaries of social media. One of the social media used by Indonesian people is Twitter. As we know, Twitter changed to X, but we are still using the name Twitter because the research was done when the name was Twitter. There are often crimes that can be detrimental. One of the crimes is cyberbullying, which can lead to depression and suicide. One of the solutions to prevent cyberbullying on Twitter is to create a Classification model using K-Nearest Neighbor (KNN) and train the model to be able to classify tweets. This research begins with collecting tweets as data using the Twitter API using the bullying keyword. The data goes through a data labeling process, data preprocessing, data weighting using TF-IDF, and Classification model using the K-Nearest Neighbor (KNN) algorithm. The best modeling results are obtained using the ratio of training data and test data of 90% and 10%, with a k value of 13. The results in the "bullying" class obtained a recall of 0.923, a precision of 0.960, an f1-score of 0.941, while in the class "not bullying" obtained a recall of 0.958, a precision of 0.920, an f1-score of 0.939, with an accuracy of 0.940.

Pneumonia is a severe lung infection that can be fatal if not detected early. This study aims to develop an effective deep-learning model for detecting pneumonia from chest X-ray images by 5,863 Paul Mooney's images from the Kaggle dataset; this research implements three deep-learning architectures: a basic Convolutional Neural Network (CNN), VGG16, and ResNet50V2. The model development process includes data preprocessing, image augmentation, and transfer learning techniques. Results show significant performance improvement, with the basic CNN model initially achieving 80.60% accuracy and the combined ResNet50V2 model reaching 92.03% accuracy. Analysis using a confusion matrix on the best model produced 93.30% precision, 94.0% recall, and 93.64% F1-score. This study contributes to the development of AI-based pneumonia diagnostic tools, potentially improving the accuracy and speed of disease detection in clinical practice.