IDF-Sign: Addressing Inconsistent Depth Features for Dynamic Sign Word Recognition

A TinyDL Model for Gesture-Based Air Handwriting Arabic Numbers and Simple Arabic Letters Recognition

“…Lastly, Khan et al focused on improving recognition through attention-based sequenceto-sequence models, demonstrating the potential for enhanced accuracy in complex gesture recognition tasks [48]. In the realm of sign language recognition, Abdullahi et al addressed the challenges of inconsistent depth features with their IDF-Sign model, showing promising results in dynamic sign word recognition [40]. Additionally, Abdullahi et al proposed a spatial-temporal feature-based end-to-end Fourier network for 3D sign language recognition, further advancing the field [49].…”

“…Gesture recognition technology [15], [16], [41], [42] is at the forefront of this paradigm shift, allowing users to interact with machines through simple movements [40]. Despite its promise, the technology faces challenges in accuracy, responsiveness, and adaptability to diverse environments.…”

Enhanced Weak Spatial Modeling Through CNN-Based Deep Sign Language Skeletal Feature Transformation

“…Sign language is the movement of the hand and/or body to transcribe an action that can be understood by the native speaker (native speaker means a person with a hard of hearing) or a professional signer (an expert with knowledge of sign language) [3]. There are over 400 sign languages [4], [5] such as Arabic sign language, American sign language, Greek sign language, Hausa sign language, etc. Because of this, a unified Arabic Sign Language (ArSL) is promulgated by the League of Arab States for formal adaptation across the Arab world [6].…”

“…Skeletal hand features are also robust to variations in background, lighting, and distance [3], which can affect the performance of video-based methods. Skeletal hand features can be obtained from sensor devices [4], such as depth cameras, or hand pose estimators, which use deep learning to detect and track the key points of the hand skeleton from images or videos. Some sign language models use skeleton-based methods alone, while others combine them with other modalities, such as RGB and depth, to provide complementary information [13].…”

“… Dataset link3 KArSL-502 Dataset link4 Depth and Leap motion Dataset linkAlgorithm 1 SCT algorithm 1: start 2: set C ← (n, f ) {feature matrix with n instances and f features} 3: set C f {feature matrix consists of only subset f } 4: set l {class variable} 5: set W {power set of the feature sets} 6: set M (•) {Performance metric} 7: set F {present subset of coordinates being considered} 8: set M f {metric evaluation on selected subsets} 9: set P (•) {function that computes a statistic for each feature} 10: set θ {Threshold value} 11: set T h {Threshold object} 12: output F 13: repeat 14:…”