Recently every scenario involving authentication and identification, biometric security assurance is essential. Due to eye iris stable and remarkable texture variation, iris recognition is thought to be the most trustworthy biometric recognition and iris recognition is used to identify those who need a high level of security. Generally, seven phases are included in recognition system: the acquisition phase, the preprocessing phase, the segmentation phase, the normalization phase and finally feature extraction phase. The development of artificial intelligence offers a fantastic potential to expand the use of iris recognition for protecting people's personal information and convolutional neural networks are a useful method that are well suited for pattern recognition and image processing because of their efficiency and adaptability. The purpose of current study is to develop a high-accuracy system for person identification from eye iris using a convolutional neural network CNN. The suggested method has been effectively put into practice by using three datasets. The deep recognition algorithm is trained using iris samples from these datasets that contain 50 different people, including data from both eyes. The results amply the effectiveness of the experimental evaluation on iris images especially from the AMF dataset, where the trained model was successful in achieving 98.46% testing accuracy.