Hand sign recognition is a vital technology in the human-computer interaction, enabling individuals to communicate with machines naturally and effectively. An innovative approach for real-time hand sign identification with the help of CNN and OpenCV is introduced with the fusion of computer vision and deep learning that can accurately interpret and classify an extensive range of hand signs and gestures. This research contributes significantly to the fields of computer vision and human-computer interaction, offering a practical and efficient solution for hand sign recognition. The combination of CNN and OpenCV presents a promising avenue for enhancing accessibility and communication, especially in environments where verbal communication is limited or non-existent. The model is trained with multiple data so that the system can recognize the hand gestures more precisely. Pre-trained architectures like ResNet and MobileNet are combined with the CNN model using ensemble learning and the performance is improved when compared to all the three CNN architectures individually. The ensemble model provides better accuracy of 96 %. The potential applications of this technology are vast, from assisting the hearing-impaired in understanding sign language to more immersive and intuitive interactions. Overall, the approach holds the promise of bridging the gap between human gestures and machine understanding, opening new doors for meaningful interactions between individuals and intelligent systems.