The bodily motion of human hand generates different gestures which are considered as non-verbal communication to convey a message to receivers. Human hand gestures have wide real-time applications in different areas, i.e., human-computer interactions (HCI), robot automation, household control, gaming, medical fields, and a major way of communication between deaf peoples. Various novel and innovative methods are used for hand gestures identification and recognition. In this paper, a systematic approach based on deep convolutional neural network (CNN) is developed for the identification and recognition of human hand gestures. The process pipeline of the proposed approach is a) segmentation of hand’s region of interest from image b) features extraction c) recognition of gestures based on CNN classifier. The proposed method is tested and verified on open access hand dataset. Experimental results validate the effectiveness and robustness of our proposed approach in terms of hand gestures recognition. Moreover, our methods achieved better recognition performance and outperformed well-known state-of-the-art gestures recognition methods.