Professional Documents
Culture Documents
Alphanumeric Recognition Using Hand Gestures: Shashank Krishna Naik, Mihir Singh, Pratik Goswami, Mahadeva Swamy GN
Alphanumeric Recognition Using Hand Gestures: Shashank Krishna Naik, Mihir Singh, Pratik Goswami, Mahadeva Swamy GN
7. Feature selection:
a. Feature selection is the process of reducing the number of input variables when developing a predictive
model.
b. It is desirable to reduce the number of input variables to both reduce the computational cost of modeling
and, in some cases, to improve the performance of the model.
c. Before feature selection.
['0',’1’,'2', '3', '4',’5’, '6', '7', '8',’9’, '10', '11', '12',’13’, '14', '15', '16',’17’, '18', '19', '20']
d. After feature selection.
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[2045]
e-ISSN: 2582-520
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:04/Issue:06/June-2022 Impact Factor- 6.752 www.irjmets.com
['0','2', '3', '4', '6', '7', '8', '10', '11', '12', '14', '15', '16', '18', '19', '20']
[Note : ‘0’ represents the hand landmark distance from 9 to 0 as shown in fig. 1. Similarly all the feature
represent the euclidean distance from hand landmark point ‘9’]
8. When we press the capture button the calculated distance will be sent to the backend for processing.
9. Then these distances are passed through a supervised learning algorithm (KNN) for multi-class
classification to classify data or predict alphabet or numbers accurately.
10. The predicted data is sent as response to the front-end application and displayed on screen.
In such a manner we plan to process the sequence of captured images and identify individual alphabets or a
number, by using these series of alphabets we can form a meaningful sentence.
IV. SYSTEM ARCHITECTURE
Recall : Percentage of positive instances out of the total actual positive instances.
Performance of KNN model is high for our implementation because our implementation has more features to recognise
or classify and it is proven that in these conditions KNN accuracy is better than SVM. So, we chose the KNN classifier.
b. Value of K in KNN classifier
To determine the optimal value of K we calculated the mean error for K value from one to forty. And which one
has the lowest mean error is considered as K value.