Sign Language Recognition (SLR) has an important role among the deaf-dump community since it is used as a medium of instruction to execute daily activities such as communication, teaching, learning, and social interactions. In this paper, a real-time model has been implemented for Kurdish sign recognition using Convolutional Neural Network (CNN) algorithm. The main objective of this study is to recognize the Kurdish alphabetic. The model has been trained and predicted on the KuSL2022 dataset using different activation functions for a number of epochs. The dataset consists of 71,400 images for the 34 Kurdish sign languages and alphabets collected from two different datasets. The accuracy of the proposed method is evaluated on a dataset of real images collected from many users. The obtained results show that the proposed system's performance increased for both classification and prediction models, with an average train accuracy of 99.91 %. These results outperform previous studies on Kurdish sign language in term of accuracy detection and recognition.