ИСТИНА

This paper proposes a gesture recognition approach in which morphological and trajectories are analyzed. A video input containing the gesture demonstration is given to the algorithm as an input and the analysis are carried out frame by frame in 3 stages. At the stage of primary processing, the location and shape of the face and hands are identified. Secondary processing includes extracting information about the change in these parameters from the video recording. At the decision making stage, a two-step algorithm is used based on a comparison of the motion trajectories and key object shapes between the test gesture and the reference gestures. The experiments were performed on selected classes from UOM-SL2020 sign language data set and a high recognition F1-score of 0.8 was achieved.