Challenge: To make a module to identify Indian Sign Language's sign for A,B,C,D
Features List: 1)Pinch Strength 2)Grab Strength 3)Thumb pointing direction x axis 4)Thumb pointing direction y axis 5)Thumb pointing direction z axis 6)Index pointing direction x axis . . 17)Pinky pointing direction z axis 18)Angle between metacarpal bone and proximal bone of thumb 19)Angle between proximal bone and intermediate bone of thumb 20)Angle between metacarpal bone and proximal bone of index finger 21)Angle between proximal bone and intermediate bone of index finger . . 27)Angle between proximal bone and intermediate bone of pinky 28)Angle between thumb and index 29)Angle between index and middle 30)Angle between middle and ring 31)Angle between ring and pinky
label list: A,B,C,D
Different accuracy lists for different Classifiers: (Models were trained combinedly from Lalu, Sai and Sandy's hand gestures and test data was collected individually from Rishav,Ashutosh,Adwesh,Sripad)
KNNeighbours Classifier: 100% accuracy for Rishav's Data(Test) 93.1% accuracy for Asutosh's Data(Test1) 100% accuracy for Adwesh's Data (Test2) 98.27% accuracy for Sripad's Data(Test3)
SVM Classifier: 100% accuracy for Rishav's Data(Test) 88.5% accuracy for Asutosh's Data(Test1) 98.8% accuracy for Adwesh's Data (Test2) 88.1% accuracy for Sripad's Data(Test3)
DecisionTree Classifier:
74.83% accuracy for Rishav's Data(Test) 99.93% accuracy for Asutosh's Data(Test1) 73.74% accuracy for Adwesh's Data (Test2) 73.88% accuracy for Sripad's Data(Test3)
Stochastic Gradient Descent: 76.6% accuracy for Rishav's Data(Test) 74.23% accuracy for Asutosh's Data(Test1) 74.89% accuracy for Adwesh's Data (Test2) 74.65% accuracy for Sripad's Data(Test3)