This study focuses on real-time hand gesture recognition in Turkish sign language detection using the YOLOv4-CSP algorithm, a state-of-the-art object detection model based on convolutional neural networks (CNNs). The proposed method optimizes the YOLOv4-CSP algorithm by integrating CSPNet into the neck of the network to enhance learning capabilities and performance. The model uses the Mish activation function, CIoU loss function, and transformer blocks to improve detection efficiency. The dataset consists of 1500 labeled images of Turkish sign language numbers, which are trained using transfer learning techniques. The proposed YOLOv4-CSP model achieves high precision, recall, F1 score, and mAP values (98.95%, 98.15%, 98.55%, and 99.49%, respectively) with a processing time of 9.8 ms. The model outperforms previous YOLO series models in terms of both real-time performance and accurate hand sign prediction, regardless of background complexity. The study demonstrates the effectiveness of the proposed method in real-time Turkish sign language recognition, making it a valuable reference for future research in this field.This study focuses on real-time hand gesture recognition in Turkish sign language detection using the YOLOv4-CSP algorithm, a state-of-the-art object detection model based on convolutional neural networks (CNNs). The proposed method optimizes the YOLOv4-CSP algorithm by integrating CSPNet into the neck of the network to enhance learning capabilities and performance. The model uses the Mish activation function, CIoU loss function, and transformer blocks to improve detection efficiency. The dataset consists of 1500 labeled images of Turkish sign language numbers, which are trained using transfer learning techniques. The proposed YOLOv4-CSP model achieves high precision, recall, F1 score, and mAP values (98.95%, 98.15%, 98.55%, and 99.49%, respectively) with a processing time of 9.8 ms. The model outperforms previous YOLO series models in terms of both real-time performance and accurate hand sign prediction, regardless of background complexity. The study demonstrates the effectiveness of the proposed method in real-time Turkish sign language recognition, making it a valuable reference for future research in this field.