Hand Posture Recognition on Photographic Images using YOLO Algorithm in Comparison with Convolution Neural Network to Improve Accuracy

Abstract

Aim: The main aim of the research work is to compare the performance of a novel YOLO algorithm in comparison with Convolution Neural Networks (CNN) to improve accuracy. Materials and Methods: In this research, Group 1 is considered as the YOLO algorithm and Group 2 is considered as CNN, recognizing the YOLO algorithm for public use when compared to CNN with higher significance value with 20 samples each group consists of 10 samples. The G power is 0.8, the alpha is 0.05, the beta is 0.2. Results: The accuracy of YOLO is significantly improved than CNN and there is a statistical significance observed as 0.0049 <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\mathbf{p} &lt; \boldsymbol{0.05})$</tex> . The accuracy for Group 1 is 95.66%, for Group 2 is 94.76%. Conclusion: YOLO algorithm is significantly more accurate compared to the convolution neural network.