Sinhala Sign Language Interpreter Optimized for Real – Time Implementation on a Mobile Device

Advancements in Vision-Based Sign Language Recognition: A Comprehensive Review

“…After the data-capturing stage [15], Dhanawansa and Rajakaruna used a method to identify the important frames from the captured frame sequence. They used a 3-frame difference method to extract signs from the frame sequence.…”

“…It is the process of isolating and delineating an image's portion corresponding to the human skin. In research [15], they used skin segmentation and subsequent filtering to isolate the critical regions of the skin within the field of view using the YCbCr colour scheme. The features of the resulting binary mask were preserved through the application of morphological transforms.…”

“…Subsequently, Morphological filtering was done to repair and smooth the segmented image. Similar to [15], Peiris [7] used a skin detection method where they used YCbCr as the colour scheme of choice to detect skin rather than RGB. They created a mask to identify the outlines of the hand in the next stage.…”

“…In [15], Dhanawansa and Rajakaruna utilized a combination of hand detection and tracking algorithms, binary masking, and template matching to identify the gestures in the image effectively. Dissanayake et al [21] performed edge detection using a thresholding method.…”

“…In [15], images are input to a CNN model. The prediction depends on 3 consecutive frames of a sign, and 2/3 majority is considered.…”