Hand Gesture Recognition via Lightweight VGG16 and Ensemble Classifier

Gesture recognition has been studied for a while within the fields of computer vision and pattern recognition. A gesture can be defined as meaningful physical movement fingers, hands, arms, or other parts body with purpose to convey information environment interaction. For instance, hand (HGR) used recognize sign language which is primary means communication by deaf mute. Vision-based HGR critical in its application; however, there are challenges that will need overcome such variations background, illuminations, orientation size similarities among gestures. The traditional machine learning approach widely vision-based recent years but complexity processing major challenge—especially on handcrafted feature extraction. effectiveness extraction technique was not proven across various datasets comparison deep techniques. Therefore, hybrid network architecture dubbed Lightweight VGG16 Random Forest (Lightweight VGG16-RF) proposed model adopts techniques via convolutional neural (CNN) using method perform classification. Experiments were carried out publicly available American Sign Language (ASL), ASL Digits NUS Hand Posture dataset. experimental results demonstrate model, combination lightweight random forest, outperforms methods.