2022 7th International Conference on Communication and Electronics Systems (ICCES) · 2022 · 15 citations · 0 references
ComputingComputer ScienceComputer VisionFeature EngineeringMachine LearningFeature DetectionFeature ExtractionFeature SelectionKernel MethodsPattern RecognitionSupport Vector MachinesCanny Edge DetectorRotated BriefScale-invariant Feature TransformScale-invariant Feature Transform (Sift)Sign Language RecognitionSupport Vector ClassificationSupport Vector Machine ClassificationVisual Feature Extraction
Sign language recognition is becoming crucial these days. To bring down the communication barrier between speech and hearing-impaired people automatic sign language recognition systems are being constantly developed and enhanced for accuracy. In this paper, we are going to compare the accuracy of two feature extraction techniques namely Oriented FAST and Rotated BRIEF (ORB) and Scale Invariant Feature Transform (SIFT). The novel model for the system is canny edge detection, feature extraction (ORB vs. SIFT), and Bag of Words technique. Because the controlled environment dataset is unfeasible, a new dataset for Indian Sign Language is produced in order to find the optimum feature extraction approach in an uncontrolled setting. Support Vector Machine (SVM) is used as a classification method.