Automatic Gesture Recognizer using Motion Tracking Device and Support Vector Machine

The proposed model describes a system consisting of hardware + software recognizing the character gestures made in free space and interpret the corresponding result. Hardware devices used are NodeMCU, ESP8266 and MPU9250. MPU9250 acts as a sensor and NodeMCU acts as a controller in the system. Thus, helps in acquiring data of each alphabet, digit, and special character gestures. Data generated by hardware will be in numeric form and this numeric data is further used for analysis. The classification carried out by algorithms namely are KNN (K-Nearest Neighbours), Decision-Tree, SVM (Support Vector Machine). The performance measures are calculated using accuracy, precision, recall, and F1-Score. The comparative analysis shows that the SVM algorithm with linear Kernel provides better results with an accuracy of 91.66%, precision 92.68%, recall 92.49%, and F1-Score 92.11%.